🕷️ Crawler Inspector

URL Lookup

Direct Parameter Lookup

Raw Queries and Responses

1. Shard Calculation

Query:
Response:
Calculated Shard: 59 (from laksa156)

2. Crawled Status Check

Query:
Response:

3. Robots.txt Check

Query:
Response:

4. Spam/Ban Check

Query:
Response:

5. Seen Status Check

ℹ️ Skipped - page is already crawled

📄
INDEXABLE
✅
CRAWLED
15 hours ago
🤖
ROBOTS ALLOWED

Page Info Filters

FilterStatusConditionDetails
HTTP statusPASSdownload_http_code = 200HTTP 200
Age cutoffPASSdownload_stamp > now() - 6 MONTH0 months ago
History dropPASSisNull(history_drop_reason)No drop reason
Spam/banPASSfh_dont_index != 1 AND ml_spam_score = 0ml_spam_score=0
CanonicalPASSmeta_canonical IS NULL OR = '' OR = src_unparsedNot set

Page Details

PropertyValue
URLhttps://pypi.org/project/openai/
Last Crawled2026-04-09 18:23:03 (15 hours ago)
First Indexed2018-05-03 19:54:03 (7 years ago)
HTTP Status Code200
Meta Titleopenai · PyPI
Meta DescriptionThe official Python library for the openai API
Meta Canonicalnull
Boilerpipe Text
OpenAI Python API library The OpenAI Python library provides convenient access to the OpenAI REST API from any Python 3.9+ application. The library includes type definitions for all request params and response fields, and offers both synchronous and asynchronous clients powered by httpx . It is generated from our OpenAPI specification with Stainless . Documentation The REST API documentation can be found on platform.openai.com . The full API of this library can be found in api.md . Installation # install from PyPI pip install openai Usage The full API of this library can be found in api.md . The primary API for interacting with OpenAI models is the Responses API . You can generate text from the model with the code below. import os from openai import OpenAI client = OpenAI ( # This is the default and can be omitted api_key = os . environ . get ( "OPENAI_API_KEY" ), ) response = client . responses . create ( model = "gpt-5.2" , instructions = "You are a coding assistant that talks like a pirate." , input = "How do I check if a Python object is an instance of a class?" , ) print ( response . output_text ) The previous standard (supported indefinitely) for generating text is the Chat Completions API . You can use that API to generate text from the model with the code below. from openai import OpenAI client = OpenAI () completion = client . chat . completions . create ( model = "gpt-5.2" , messages = [ { "role" : "developer" , "content" : "Talk like a pirate." }, { "role" : "user" , "content" : "How do I check if a Python object is an instance of a class?" , }, ], ) print ( completion . choices [ 0 ] . message . content ) While you can provide an api_key keyword argument, we recommend using python-dotenv to add OPENAI_API_KEY="My API Key" to your .env file so that your API key is not stored in source control. Get an API key here . Workload Identity Authentication For secure, automated environments like cloud-managed Kubernetes, Azure, and Google Cloud Platform, you can use workload identity authentication with short-lived tokens from cloud identity providers instead of long-lived API keys. Kubernetes (service account tokens) from openai import OpenAI from openai.auth import k8s_service_account_token_provider client = OpenAI ( workload_identity = { "client_id" : "your-client-id" , "identity_provider_id" : "idp-123" , "service_account_id" : "sa-456" , "provider" : k8s_service_account_token_provider ( "/var/run/secrets/kubernetes.io/serviceaccount/token" ), }, organization = "org-xyz" , project = "proj-abc" , ) response = client . chat . completions . create ( model = "gpt-4" , messages = [{ "role" : "user" , "content" : "Hello!" }], ) Azure (managed identity) from openai import OpenAI from openai.auth import azure_managed_identity_token_provider client = OpenAI ( workload_identity = { "client_id" : "your-client-id" , "identity_provider_id" : "idp-123" , "service_account_id" : "sa-456" , "provider" : azure_managed_identity_token_provider ( resource = "https://management.azure.com/" , ), }, ) Google Cloud Platform (compute engine metadata) from openai import OpenAI from openai.auth import gcp_id_token_provider client = OpenAI ( workload_identity = { "client_id" : "your-client-id" , "identity_provider_id" : "idp-123" , "service_account_id" : "sa-456" , "provider" : gcp_id_token_provider ( audience = "https://api.openai.com/v1" ), }, ) Custom subject token provider from openai import OpenAI def get_custom_token () -> str : return "your-jwt-token" client = OpenAI ( workload_identity = { "client_id" : "your-client-id" , "identity_provider_id" : "idp-123" , "service_account_id" : "sa-456" , "provider" : { "token_type" : "jwt" , "get_token" : get_custom_token , }, } ) You can also customize the token refresh buffer (default is 1200 seconds (20 minutes) before expiration): from openai import OpenAI from openai.auth import k8s_service_account_token_provider client = OpenAI ( workload_identity = { "client_id" : "your-client-id" , "identity_provider_id" : "idp-123" , "service_account_id" : "sa-456" , "provider" : k8s_service_account_token_provider ( "/var/token" ), "refresh_buffer_seconds" : 120.0 , } ) Vision With an image URL: prompt = "What is in this image?" img_url = "https://upload.wikimedia.org/wikipedia/commons/thumb/d/d5/2023_06_08_Raccoon1.jpg/1599px-2023_06_08_Raccoon1.jpg" response = client . responses . create ( model = "gpt-5.2" , input = [ { "role" : "user" , "content" : [ { "type" : "input_text" , "text" : prompt }, { "type" : "input_image" , "image_url" : f " { img_url } " }, ], } ], ) With the image as a base64 encoded string: import base64 from openai import OpenAI client = OpenAI () prompt = "What is in this image?" with open ( "path/to/image.png" , "rb" ) as image_file : b64_image = base64 . b64encode ( image_file . read ()) . decode ( "utf-8" ) response = client . responses . create ( model = "gpt-5.2" , input = [ { "role" : "user" , "content" : [ { "type" : "input_text" , "text" : prompt }, { "type" : "input_image" , "image_url" : f "data:image/png;base64, { b64_image } " }, ], } ], ) Async usage Simply import AsyncOpenAI instead of OpenAI and use await with each API call: import os import asyncio from openai import AsyncOpenAI client = AsyncOpenAI ( # This is the default and can be omitted api_key = os . environ . get ( "OPENAI_API_KEY" ), ) async def main () -> None : response = await client . responses . create ( model = "gpt-5.2" , input = "Explain disestablishmentarianism to a smart five year old." ) print ( response . output_text ) asyncio . run ( main ()) Functionality between the synchronous and asynchronous clients is otherwise identical. With aiohttp By default, the async client uses httpx for HTTP requests. However, for improved concurrency performance you may also use aiohttp as the HTTP backend. You can enable this by installing aiohttp : # install from PyPI pip install openai [ aiohttp ] Then you can enable it by instantiating the client with http_client=DefaultAioHttpClient() : import os import asyncio from openai import DefaultAioHttpClient from openai import AsyncOpenAI async def main () -> None : async with AsyncOpenAI ( api_key = os . environ . get ( "OPENAI_API_KEY" ), # This is the default and can be omitted http_client = DefaultAioHttpClient (), ) as client : chat_completion = await client . chat . completions . create ( messages = [ { "role" : "user" , "content" : "Say this is a test" , } ], model = "gpt-5.2" , ) asyncio . run ( main ()) Streaming responses We provide support for streaming responses using Server Side Events (SSE). from openai import OpenAI client = OpenAI () stream = client . responses . create ( model = "gpt-5.2" , input = "Write a one-sentence bedtime story about a unicorn." , stream = True , ) for event in stream : print ( event ) The async client uses the exact same interface. import asyncio from openai import AsyncOpenAI client = AsyncOpenAI () async def main (): stream = await client . responses . create ( model = "gpt-5.2" , input = "Write a one-sentence bedtime story about a unicorn." , stream = True , ) async for event in stream : print ( event ) asyncio . run ( main ()) Realtime API The Realtime API enables you to build low-latency, multi-modal conversational experiences. It currently supports text and audio as both input and output, as well as function calling through a WebSocket connection. Under the hood the SDK uses the websockets library to manage connections. The Realtime API works through a combination of client-sent events and server-sent events. Clients can send events to do things like update session configuration or send text and audio inputs. Server events confirm when audio responses have completed, or when a text response from the model has been received. A full event reference can be found here and a guide can be found here . Basic text based example: import asyncio from openai import AsyncOpenAI async def main (): client = AsyncOpenAI () async with client . realtime . connect ( model = "gpt-realtime" ) as connection : await connection . session . update ( session = { "type" : "realtime" , "output_modalities" : [ "text" ]} ) await connection . conversation . item . create ( item = { "type" : "message" , "role" : "user" , "content" : [{ "type" : "input_text" , "text" : "Say hello!" }], } ) await connection . response . create () async for event in connection : if event . type == "response.output_text.delta" : print ( event . delta , flush = True , end = "" ) elif event . type == "response.output_text.done" : print () elif event . type == "response.done" : break asyncio . run ( main ()) However the real magic of the Realtime API is handling audio inputs / outputs, see this example TUI script for a fully fledged example. Realtime error handling Whenever an error occurs, the Realtime API will send an error event and the connection will stay open and remain usable. This means you need to handle it yourself, as no errors are raised directly by the SDK when an error event comes in. client = AsyncOpenAI () async with client . realtime . connect ( model = "gpt-realtime" ) as connection : ... async for event in connection : if event . type == 'error' : print ( event . error . type ) print ( event . error . code ) print ( event . error . event_id ) print ( event . error . message ) Using types Nested request parameters are TypedDicts . Responses are Pydantic models which also provide helper methods for things like: Serializing back into JSON, model.to_json() Converting to a dictionary, model.to_dict() Typed requests and responses provide autocomplete and documentation within your editor. If you would like to see type errors in VS Code to help catch bugs earlier, set python.analysis.typeCheckingMode to basic . Pagination List methods in the OpenAI API are paginated. This library provides auto-paginating iterators with each list response, so you do not have to request successive pages manually: from openai import OpenAI client = OpenAI () all_jobs = [] # Automatically fetches more pages as needed. for job in client . fine_tuning . jobs . list ( limit = 20 , ): # Do something with job here all_jobs . append ( job ) print ( all_jobs ) Or, asynchronously: import asyncio from openai import AsyncOpenAI client = AsyncOpenAI () async def main () -> None : all_jobs = [] # Iterate through items across all pages, issuing requests as needed. async for job in client . fine_tuning . jobs . list ( limit = 20 , ): all_jobs . append ( job ) print ( all_jobs ) asyncio . run ( main ()) Alternatively, you can use the .has_next_page() , .next_page_info() , or .get_next_page() methods for more granular control working with pages: first_page = await client . fine_tuning . jobs . list ( limit = 20 , ) if first_page . has_next_page (): print ( f "will fetch next page using these details: { first_page . next_page_info () } " ) next_page = await first_page . get_next_page () print ( f "number of items we just fetched: { len ( next_page . data ) } " ) # Remove `await` for non-async usage. Or just work directly with the returned data: first_page = await client . fine_tuning . jobs . list ( limit = 20 , ) print ( f "next page cursor: { first_page . after } " ) # => "next page cursor: ..." for job in first_page . data : print ( job . id ) # Remove `await` for non-async usage. Nested params Nested parameters are dictionaries, typed using TypedDict , for example: from openai import OpenAI client = OpenAI () response = client . chat . responses . create ( input = [ { "role" : "user" , "content" : "How much ?" , } ], model = "gpt-5.2" , response_format = { "type" : "json_object" }, ) File uploads Request parameters that correspond to file uploads can be passed as bytes , or a PathLike instance or a tuple of (filename, contents, media type) . from pathlib import Path from openai import OpenAI client = OpenAI () client . files . create ( file = Path ( "input.jsonl" ), purpose = "fine-tune" , ) The async client uses the exact same interface. If you pass a PathLike instance, the file contents will be read asynchronously automatically. Webhook Verification Verifying webhook signatures is optional but encouraged . For more information about webhooks, see the API docs . Parsing webhook payloads For most use cases, you will likely want to verify the webhook and parse the payload at the same time. To achieve this, we provide the method client.webhooks.unwrap() , which parses a webhook request and verifies that it was sent by OpenAI. This method will raise an error if the signature is invalid. Note that the body parameter must be the raw JSON string sent from the server (do not parse it first). The .unwrap() method will parse this JSON for you into an event object after verifying the webhook was sent from OpenAI. from openai import OpenAI from flask import Flask , request app = Flask ( __name__ ) client = OpenAI () # OPENAI_WEBHOOK_SECRET environment variable is used by default @app . route ( "/webhook" , methods = [ "POST" ]) def webhook (): request_body = request . get_data ( as_text = True ) try : event = client . webhooks . unwrap ( request_body , request . headers ) if event . type == "response.completed" : print ( "Response completed:" , event . data ) elif event . type == "response.failed" : print ( "Response failed:" , event . data ) else : print ( "Unhandled event type:" , event . type ) return "ok" except Exception as e : print ( "Invalid signature:" , e ) return "Invalid signature" , 400 if __name__ == "__main__" : app . run ( port = 8000 ) Verifying webhook payloads directly In some cases, you may want to verify the webhook separately from parsing the payload. If you prefer to handle these steps separately, we provide the method client.webhooks.verify_signature() to only verify the signature of a webhook request. Like .unwrap() , this method will raise an error if the signature is invalid. Note that the body parameter must be the raw JSON string sent from the server (do not parse it first). You will then need to parse the body after verifying the signature. import json from openai import OpenAI from flask import Flask , request app = Flask ( __name__ ) client = OpenAI () # OPENAI_WEBHOOK_SECRET environment variable is used by default @app . route ( "/webhook" , methods = [ "POST" ]) def webhook (): request_body = request . get_data ( as_text = True ) try : client . webhooks . verify_signature ( request_body , request . headers ) # Parse the body after verification event = json . loads ( request_body ) print ( "Verified event:" , event ) return "ok" except Exception as e : print ( "Invalid signature:" , e ) return "Invalid signature" , 400 if __name__ == "__main__" : app . run ( port = 8000 ) Handling errors When the library is unable to connect to the API (for example, due to network connection problems or a timeout), a subclass of openai.APIConnectionError is raised. When the API returns a non-success status code (that is, 4xx or 5xx response), a subclass of openai.APIStatusError is raised, containing status_code and response properties. All errors inherit from openai.APIError . import openai from openai import OpenAI client = OpenAI () try : client . fine_tuning . jobs . create ( model = "gpt-4o" , training_file = "file-abc123" , ) except openai . APIConnectionError as e : print ( "The server could not be reached" ) print ( e . __cause__ ) # an underlying Exception, likely raised within httpx. except openai . RateLimitError as e : print ( "A 429 status code was received; we should back off a bit." ) except openai . APIStatusError as e : print ( "Another non-200-range status code was received" ) print ( e . status_code ) print ( e . response ) Error codes are as follows: Status Code Error Type 400 BadRequestError 401 AuthenticationError 403 PermissionDeniedError 404 NotFoundError 422 UnprocessableEntityError 429 RateLimitError >=500 InternalServerError N/A APIConnectionError Request IDs For more information on debugging requests, see these docs All object responses in the SDK provide a _request_id property which is added from the x-request-id response header so that you can quickly log failing requests and report them back to OpenAI. response = await client . responses . create ( model = "gpt-5.2" , input = "Say 'this is a test'." , ) print ( response . _request_id ) # req_123 Note that unlike other properties that use an _ prefix, the _request_id property is public. Unless documented otherwise, all other _ prefix properties, methods and modules are private . [!IMPORTANT] If you need to access request IDs for failed requests you must catch the APIStatusError exception import openai try : completion = await client . chat . completions . create ( messages = [{ "role" : "user" , "content" : "Say this is a test" }], model = "gpt-5.2" ) except openai . APIStatusError as exc : print ( exc . request_id ) # req_123 raise exc Retries Certain errors are automatically retried 2 times by default, with a short exponential backoff. Connection errors (for example, due to a network connectivity problem), 408 Request Timeout, 409 Conflict, 429 Rate Limit, and >=500 Internal errors are all retried by default. You can use the max_retries option to configure or disable retry settings: from openai import OpenAI # Configure the default for all requests: client = OpenAI ( # default is 2 max_retries = 0 , ) # Or, configure per-request: client . with_options ( max_retries = 5 ) . chat . completions . create ( messages = [ { "role" : "user" , "content" : "How can I get the name of the current day in JavaScript?" , } ], model = "gpt-5.2" , ) Timeouts By default requests time out after 10 minutes. You can configure this with a timeout option, which accepts a float or an httpx.Timeout object: from openai import OpenAI # Configure the default for all requests: client = OpenAI ( # 20 seconds (default is 10 minutes) timeout = 20.0 , ) # More granular control: client = OpenAI ( timeout = httpx . Timeout ( 60.0 , read = 5.0 , write = 10.0 , connect = 2.0 ), ) # Override per-request: client . with_options ( timeout = 5.0 ) . chat . completions . create ( messages = [ { "role" : "user" , "content" : "How can I list all files in a directory using Python?" , } ], model = "gpt-5.2" , ) On timeout, an APITimeoutError is thrown. Note that requests that time out are retried twice by default . Advanced Logging We use the standard library logging module. You can enable logging by setting the environment variable OPENAI_LOG to info . $ export OPENAI_LOG = info Or to debug for more verbose logging. How to tell whether None means null or missing In an API response, a field may be explicitly null , or missing entirely; in either case, its value is None in this library. You can differentiate the two cases with .model_fields_set : if response . my_field is None : if 'my_field' not in response . model_fields_set : print ( 'Got json like {} , without a "my_field" key present at all.' ) else : print ( 'Got json like {"my_field": null}.' ) Accessing raw response data (e.g. headers) The "raw" Response object can be accessed by prefixing .with_raw_response. to any HTTP method call, e.g., from openai import OpenAI client = OpenAI () response = client . chat . completions . with_raw_response . create ( messages = [{ "role" : "user" , "content" : "Say this is a test" , }], model = "gpt-5.2" , ) print ( response . headers . get ( 'X-My-Header' )) completion = response . parse () # get the object that `chat.completions.create()` would have returned print ( completion ) These methods return a LegacyAPIResponse object. This is a legacy class as we're changing it slightly in the next major version. For the sync client this will mostly be the same with the exception of content & text will be methods instead of properties. In the async client, all methods will be async. A migration script will be provided & the migration in general should be smooth. .with_streaming_response The above interface eagerly reads the full response body when you make the request, which may not always be what you want. To stream the response body, use .with_streaming_response instead, which requires a context manager and only reads the response body once you call .read() , .text() , .json() , .iter_bytes() , .iter_text() , .iter_lines() or .parse() . In the async client, these are async methods. As such, .with_streaming_response methods return a different APIResponse object, and the async client returns an AsyncAPIResponse object. with client . chat . completions . with_streaming_response . create ( messages = [ { "role" : "user" , "content" : "Say this is a test" , } ], model = "gpt-5.2" , ) as response : print ( response . headers . get ( "X-My-Header" )) for line in response . iter_lines (): print ( line ) The context manager is required so that the response will reliably be closed. Making custom/undocumented requests This library is typed for convenient access to the documented API. If you need to access undocumented endpoints, params, or response properties, the library can still be used. Undocumented endpoints To make requests to undocumented endpoints, you can make requests using client.get , client.post , and other http verbs. Options on the client will be respected (such as retries) when making this request. import httpx response = client . post ( "/foo" , cast_to = httpx . Response , body = { "my_param" : True }, ) print ( response . headers . get ( "x-foo" )) Undocumented request params If you want to explicitly send an extra param, you can do so with the extra_query , extra_body , and extra_headers request options. Undocumented response properties To access undocumented response properties, you can access the extra fields like response.unknown_prop . You can also get all the extra fields on the Pydantic model as a dict with response.model_extra . Configuring the HTTP client You can directly override the httpx client to customize it for your use case, including: Support for proxies Custom transports Additional advanced functionality import httpx from openai import OpenAI , DefaultHttpxClient client = OpenAI ( # Or use the `OPENAI_BASE_URL` env var base_url = "http://my.test.server.example.com:8083/v1" , http_client = DefaultHttpxClient ( proxy = "http://my.test.proxy.example.com" , transport = httpx . HTTPTransport ( local_address = "0.0.0.0" ), ), ) You can also customize the client on a per-request basis by using with_options() : client . with_options ( http_client = DefaultHttpxClient ( ... )) Managing HTTP resources By default the library closes underlying HTTP connections whenever the client is garbage collected . You can manually close the client using the .close() method if desired, or with a context manager that closes when exiting. from openai import OpenAI with OpenAI () as client : # make requests here ... # HTTP client is now closed Microsoft Azure OpenAI To use this library with Azure OpenAI , use the AzureOpenAI class instead of the OpenAI class. [!IMPORTANT] The Azure API shape differs from the core API shape which means that the static types for responses / params won't always be correct. from openai import AzureOpenAI # gets the API Key from environment variable AZURE_OPENAI_API_KEY client = AzureOpenAI ( # https://learn.microsoft.com/azure/ai-services/openai/reference#rest-api-versioning api_version = "2023-07-01-preview" , # https://learn.microsoft.com/azure/cognitive-services/openai/how-to/create-resource?pivots=web-portal#create-a-resource azure_endpoint = "https://example-endpoint.openai.azure.com" , ) completion = client . chat . completions . create ( model = "deployment-name" , # e.g. gpt-35-instant messages = [ { "role" : "user" , "content" : "How do I output all files in a directory using Python?" , }, ], ) print ( completion . to_json ()) In addition to the options provided in the base OpenAI client, the following options are provided: azure_endpoint (or the AZURE_OPENAI_ENDPOINT environment variable) azure_deployment api_version (or the OPENAI_API_VERSION environment variable) azure_ad_token (or the AZURE_OPENAI_AD_TOKEN environment variable) azure_ad_token_provider An example of using the client with Microsoft Entra ID (formerly known as Azure Active Directory) can be found here . Versioning This package generally follows SemVer conventions, though certain backwards-incompatible changes may be released as minor versions: Changes that only affect static types, without breaking runtime behavior. Changes to library internals which are technically public but not intended or documented for external use. (Please open a GitHub issue to let us know if you are relying on such internals.) Changes that we do not expect to impact the vast majority of users in practice. We take backwards-compatibility seriously and work hard to ensure you can rely on a smooth upgrade experience. We are keen for your feedback; please open an issue with questions, bugs, or suggestions. Determining the installed version If you've upgraded to the latest version but aren't seeing any new features you were expecting then your python environment is likely still using an older version. You can determine the version that is being used at runtime with: import openai print ( openai . __version__ ) Requirements Python 3.9 or higher. Contributing See the contributing documentation .
Markdown
[Skip to main content](https://pypi.org/project/openai/#content) Switch to mobile version Warning Some features may not work without JavaScript. Please try enabling it if you encounter problems. Join us in Long Beach, CA starting May 13, 2026. Grab your ticket and discounted hotel today before they’re gone! [REGISTER FOR PYCON US\!](https://us.pycon.org/2026/attend/information/) [![PyPI](https://pypi.org/static/images/logo-small.8998e9d1.svg)](https://pypi.org/) - [Help](https://pypi.org/help/) - [Docs](https://docs.pypi.org/) - [Sponsors](https://pypi.org/sponsors/) - [Log in](https://pypi.org/account/login/?next=https%3A%2F%2Fpypi.org%2Fproject%2Fopenai%2F) - [Register](https://pypi.org/account/register/) Menu - [Help](https://pypi.org/help/) - [Docs](https://docs.pypi.org/) - [Sponsors](https://pypi.org/sponsors/) - [Log in](https://pypi.org/account/login/?next=https%3A%2F%2Fpypi.org%2Fproject%2Fopenai%2F) - [Register](https://pypi.org/account/register/) # openai 2.31.0 pip install openai Copy PIP instructions [Latest version](https://pypi.org/project/openai/) Released: Apr 8, 2026 The official Python library for the openai API ### Navigation - [Project description](https://pypi.org/project/openai/#description) - [Release history](https://pypi.org/project/openai/#history) - [Download files](https://pypi.org/project/openai/#files) ### Verified details *These details have been [verified by PyPI](https://docs.pypi.org/project_metadata/#verified-details)* ###### Owner - [OpenAI](https://pypi.org/org/openai/) ###### Maintainers [![Avatar for ddeville from gravatar.com](https://pypi-camo.freetls.fastly.net/32f89b8b480b8a373a82e43b9846aa7f3d1347e3/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f36356430636636356336623363303534396439316630303331333932626234343f73697a653d3530) ddeville](https://pypi.org/user/ddeville/) [![Avatar for dschnurr-openai from gravatar.com](https://pypi-camo.freetls.fastly.net/d276ae7e441dadd38c03dc2cddfc79a40bdfe6f4/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f39396434623863633962623533346639653134383466393432316261343765363f73697a653d3530) dschnurr-openai](https://pypi.org/user/dschnurr-openai/) [![Avatar for gdb from gravatar.com](https://pypi-camo.freetls.fastly.net/714c5137f119ec3e27cb0fc83a0042474855d1ff/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f39643862336538633662616266666461336366353063633166663863383731653f73697a653d3530) gdb](https://pypi.org/user/gdb/) [![Avatar for hponde\_oai from gravatar.com](https://pypi-camo.freetls.fastly.net/41ade05ccfd7c00016ca9067cfa93c96fae5c47c/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f66313362623930363033383863393761383663656261323239646534383633613f73697a653d3530) hponde\_oai](https://pypi.org/user/hponde_oai/) [![Avatar for jhallard from gravatar.com](https://pypi-camo.freetls.fastly.net/3eda32bb127f92256b2848a48a6a42a085858f86/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f61343634643164613232313863663031343866323334303230353766356333323f73697a653d3530) jhallard](https://pypi.org/user/jhallard/) [![Avatar for tomerkOpenAI from gravatar.com](https://pypi-camo.freetls.fastly.net/2eeda1a7257b4a01b3a5b7c9c3f21298ab30711b/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f35396231346638626664326633303562343632373563346231373466663165643f73697a653d3530) tomerkOpenAI](https://pypi.org/user/tomerkOpenAI/) ### Unverified details *These details have **not** been verified by PyPI* ###### Project links - [Homepage](https://github.com/openai/openai-python) - [Repository](https://github.com/openai/openai-python) ###### Meta - **License:** Apache Software License (Apache-2.0) - **Author:** [OpenAI](mailto:support@openai.com) - **Requires:** Python \>=3.9 - **Provides-Extra:** `aiohttp` , `datalib` , `realtime` , `voice-helpers` ###### Classifiers - **Intended Audience** - [Developers](https://pypi.org/search/?c=Intended+Audience+%3A%3A+Developers) - **License** - [OSI Approved :: Apache Software License](https://pypi.org/search/?c=License+%3A%3A+OSI+Approved+%3A%3A+Apache+Software+License) - **Operating System** - [MacOS](https://pypi.org/search/?c=Operating+System+%3A%3A+MacOS) - [Microsoft :: Windows](https://pypi.org/search/?c=Operating+System+%3A%3A+Microsoft+%3A%3A+Windows) - [OS Independent](https://pypi.org/search/?c=Operating+System+%3A%3A+OS+Independent) - [POSIX](https://pypi.org/search/?c=Operating+System+%3A%3A+POSIX) - [POSIX :: Linux](https://pypi.org/search/?c=Operating+System+%3A%3A+POSIX+%3A%3A+Linux) - **Programming Language** - [Python :: 3.9](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.9) - [Python :: 3.10](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.10) - [Python :: 3.11](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.11) - [Python :: 3.12](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.12) - [Python :: 3.13](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.13) - [Python :: 3.14](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.14) - **Topic** - [Software Development :: Libraries :: Python Modules](https://pypi.org/search/?c=Topic+%3A%3A+Software+Development+%3A%3A+Libraries+%3A%3A+Python+Modules) - **Typing** - [Typed](https://pypi.org/search/?c=Typing+%3A%3A+Typed) [Report project as malware](https://pypi.org/project/openai/submit-malware-report/) - [Project description](https://pypi.org/project/openai/#description) - [Project details](https://pypi.org/project/openai/#data) - [Release history](https://pypi.org/project/openai/#history) - [Download files](https://pypi.org/project/openai/#files) ## Project description # OpenAI Python API library [![PyPI version](https://pypi-camo.freetls.fastly.net/594569836ed5abd55c2114c0e6795992b694fd4c/68747470733a2f2f696d672e736869656c64732e696f2f707970692f762f6f70656e61692e7376673f6c6162656c3d7079706925323028737461626c6529)](https://pypi.org/project/openai/) The OpenAI Python library provides convenient access to the OpenAI REST API from any Python 3.9+ application. The library includes type definitions for all request params and response fields, and offers both synchronous and asynchronous clients powered by [httpx](https://github.com/encode/httpx). It is generated from our [OpenAPI specification](https://github.com/openai/openai-openapi) with [Stainless](https://stainlessapi.com/). ## Documentation The REST API documentation can be found on [platform.openai.com](https://platform.openai.com/docs/api-reference). The full API of this library can be found in [api.md](https://github.com/openai/openai-python/tree/main/api.md). ## Installation ``` # install from PyPI pip install openai ``` ## Usage The full API of this library can be found in [api.md](https://github.com/openai/openai-python/tree/main/api.md). The primary API for interacting with OpenAI models is the [Responses API](https://platform.openai.com/docs/api-reference/responses). You can generate text from the model with the code below. ``` import os from openai import OpenAI client = OpenAI( # This is the default and can be omitted api_key=os.environ.get("OPENAI_API_KEY"), ) response = client.responses.create( model="gpt-5.2", instructions="You are a coding assistant that talks like a pirate.", input="How do I check if a Python object is an instance of a class?", ) print(response.output_text) ``` The previous standard (supported indefinitely) for generating text is the [Chat Completions API](https://platform.openai.com/docs/api-reference/chat). You can use that API to generate text from the model with the code below. ``` from openai import OpenAI client = OpenAI() completion = client.chat.completions.create( model="gpt-5.2", messages=[ {"role": "developer", "content": "Talk like a pirate."}, { "role": "user", "content": "How do I check if a Python object is an instance of a class?", }, ], ) print(completion.choices[0].message.content) ``` While you can provide an `api_key` keyword argument, we recommend using [python-dotenv](https://pypi.org/project/python-dotenv/) to add `OPENAI_API_KEY="My API Key"` to your `.env` file so that your API key is not stored in source control. [Get an API key here](https://platform.openai.com/settings/organization/api-keys). ### Workload Identity Authentication For secure, automated environments like cloud-managed Kubernetes, Azure, and Google Cloud Platform, you can use workload identity authentication with short-lived tokens from cloud identity providers instead of long-lived API keys. #### Kubernetes (service account tokens) ``` from openai import OpenAI from openai.auth import k8s_service_account_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": k8s_service_account_token_provider( "/var/run/secrets/kubernetes.io/serviceaccount/token" ), }, organization="org-xyz", project="proj-abc", ) response = client.chat.completions.create( model="gpt-4", messages=[{"role": "user", "content": "Hello!"}], ) ``` #### Azure (managed identity) ``` from openai import OpenAI from openai.auth import azure_managed_identity_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": azure_managed_identity_token_provider( resource="https://management.azure.com/", ), }, ) ``` #### Google Cloud Platform (compute engine metadata) ``` from openai import OpenAI from openai.auth import gcp_id_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": gcp_id_token_provider(audience="https://api.openai.com/v1"), }, ) ``` #### Custom subject token provider ``` from openai import OpenAI def get_custom_token() -> str: return "your-jwt-token" client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": { "token_type": "jwt", "get_token": get_custom_token, }, } ) ``` You can also customize the token refresh buffer (default is 1200 seconds (20 minutes) before expiration): ``` from openai import OpenAI from openai.auth import k8s_service_account_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": k8s_service_account_token_provider("/var/token"), "refresh_buffer_seconds": 120.0, } ) ``` ### Vision With an image URL: ``` prompt = "What is in this image?" img_url = "https://upload.wikimedia.org/wikipedia/commons/thumb/d/d5/2023_06_08_Raccoon1.jpg/1599px-2023_06_08_Raccoon1.jpg" response = client.responses.create( model="gpt-5.2", input=[ { "role": "user", "content": [ {"type": "input_text", "text": prompt}, {"type": "input_image", "image_url": f"{img_url}"}, ], } ], ) ``` With the image as a base64 encoded string: ``` import base64 from openai import OpenAI client = OpenAI() prompt = "What is in this image?" with open("path/to/image.png", "rb") as image_file: b64_image = base64.b64encode(image_file.read()).decode("utf-8") response = client.responses.create( model="gpt-5.2", input=[ { "role": "user", "content": [ {"type": "input_text", "text": prompt}, {"type": "input_image", "image_url": f"data:image/png;base64,{b64_image}"}, ], } ], ) ``` ## Async usage Simply import `AsyncOpenAI` instead of `OpenAI` and use `await` with each API call: ``` import os import asyncio from openai import AsyncOpenAI client = AsyncOpenAI( # This is the default and can be omitted api_key=os.environ.get("OPENAI_API_KEY"), ) async def main() -> None: response = await client.responses.create( model="gpt-5.2", input="Explain disestablishmentarianism to a smart five year old." ) print(response.output_text) asyncio.run(main()) ``` Functionality between the synchronous and asynchronous clients is otherwise identical. ### With aiohttp By default, the async client uses `httpx` for HTTP requests. However, for improved concurrency performance you may also use `aiohttp` as the HTTP backend. You can enable this by installing `aiohttp`: ``` # install from PyPI pip install openai[aiohttp] ``` Then you can enable it by instantiating the client with `http_client=DefaultAioHttpClient()`: ``` import os import asyncio from openai import DefaultAioHttpClient from openai import AsyncOpenAI async def main() -> None: async with AsyncOpenAI( api_key=os.environ.get("OPENAI_API_KEY"), # This is the default and can be omitted http_client=DefaultAioHttpClient(), ) as client: chat_completion = await client.chat.completions.create( messages=[ { "role": "user", "content": "Say this is a test", } ], model="gpt-5.2", ) asyncio.run(main()) ``` ## Streaming responses We provide support for streaming responses using Server Side Events (SSE). ``` from openai import OpenAI client = OpenAI() stream = client.responses.create( model="gpt-5.2", input="Write a one-sentence bedtime story about a unicorn.", stream=True, ) for event in stream: print(event) ``` The async client uses the exact same interface. ``` import asyncio from openai import AsyncOpenAI client = AsyncOpenAI() async def main(): stream = await client.responses.create( model="gpt-5.2", input="Write a one-sentence bedtime story about a unicorn.", stream=True, ) async for event in stream: print(event) asyncio.run(main()) ``` ## Realtime API The Realtime API enables you to build low-latency, multi-modal conversational experiences. It currently supports text and audio as both input and output, as well as [function calling](https://platform.openai.com/docs/guides/function-calling) through a WebSocket connection. Under the hood the SDK uses the [`websockets`](https://websockets.readthedocs.io/en/stable/) library to manage connections. The Realtime API works through a combination of client-sent events and server-sent events. Clients can send events to do things like update session configuration or send text and audio inputs. Server events confirm when audio responses have completed, or when a text response from the model has been received. A full event reference can be found [here](https://platform.openai.com/docs/api-reference/realtime-client-events) and a guide can be found [here](https://platform.openai.com/docs/guides/realtime). Basic text based example: ``` import asyncio from openai import AsyncOpenAI async def main(): client = AsyncOpenAI() async with client.realtime.connect(model="gpt-realtime") as connection: await connection.session.update( session={"type": "realtime", "output_modalities": ["text"]} ) await connection.conversation.item.create( item={ "type": "message", "role": "user", "content": [{"type": "input_text", "text": "Say hello!"}], } ) await connection.response.create() async for event in connection: if event.type == "response.output_text.delta": print(event.delta, flush=True, end="") elif event.type == "response.output_text.done": print() elif event.type == "response.done": break asyncio.run(main()) ``` However the real magic of the Realtime API is handling audio inputs / outputs, see this example [TUI script](https://github.com/openai/openai-python/blob/main/examples/realtime/push_to_talk_app.py) for a fully fledged example. ### Realtime error handling Whenever an error occurs, the Realtime API will send an [`error` event](https://platform.openai.com/docs/guides/realtime-model-capabilities#error-handling) and the connection will stay open and remain usable. This means you need to handle it yourself, as *no errors are raised directly* by the SDK when an `error` event comes in. ``` client = AsyncOpenAI() async with client.realtime.connect(model="gpt-realtime") as connection: ... async for event in connection: if event.type == 'error': print(event.error.type) print(event.error.code) print(event.error.event_id) print(event.error.message) ``` ## Using types Nested request parameters are [TypedDicts](https://docs.python.org/3/library/typing.html#typing.TypedDict). Responses are [Pydantic models](https://docs.pydantic.dev/) which also provide helper methods for things like: - Serializing back into JSON, `model.to_json()` - Converting to a dictionary, `model.to_dict()` Typed requests and responses provide autocomplete and documentation within your editor. If you would like to see type errors in VS Code to help catch bugs earlier, set `python.analysis.typeCheckingMode` to `basic`. ## Pagination List methods in the OpenAI API are paginated. This library provides auto-paginating iterators with each list response, so you do not have to request successive pages manually: ``` from openai import OpenAI client = OpenAI() all_jobs = [] # Automatically fetches more pages as needed. for job in client.fine_tuning.jobs.list( limit=20, ): # Do something with job here all_jobs.append(job) print(all_jobs) ``` Or, asynchronously: ``` import asyncio from openai import AsyncOpenAI client = AsyncOpenAI() async def main() -> None: all_jobs = [] # Iterate through items across all pages, issuing requests as needed. async for job in client.fine_tuning.jobs.list( limit=20, ): all_jobs.append(job) print(all_jobs) asyncio.run(main()) ``` Alternatively, you can use the `.has_next_page()`, `.next_page_info()`, or `.get_next_page()` methods for more granular control working with pages: ``` first_page = await client.fine_tuning.jobs.list( limit=20, ) if first_page.has_next_page(): print(f"will fetch next page using these details: {first_page.next_page_info()}") next_page = await first_page.get_next_page() print(f"number of items we just fetched: {len(next_page.data)}") # Remove `await` for non-async usage. ``` Or just work directly with the returned data: ``` first_page = await client.fine_tuning.jobs.list( limit=20, ) print(f"next page cursor: {first_page.after}") # => "next page cursor: ..." for job in first_page.data: print(job.id) # Remove `await` for non-async usage. ``` ## Nested params Nested parameters are dictionaries, typed using `TypedDict`, for example: ``` from openai import OpenAI client = OpenAI() response = client.chat.responses.create( input=[ { "role": "user", "content": "How much ?", } ], model="gpt-5.2", response_format={"type": "json_object"}, ) ``` ## File uploads Request parameters that correspond to file uploads can be passed as `bytes`, or a [`PathLike`](https://docs.python.org/3/library/os.html#os.PathLike) instance or a tuple of `(filename, contents, media type)`. ``` from pathlib import Path from openai import OpenAI client = OpenAI() client.files.create( file=Path("input.jsonl"), purpose="fine-tune", ) ``` The async client uses the exact same interface. If you pass a [`PathLike`](https://docs.python.org/3/library/os.html#os.PathLike) instance, the file contents will be read asynchronously automatically. ## Webhook Verification Verifying webhook signatures is *optional but encouraged*. For more information about webhooks, see [the API docs](https://platform.openai.com/docs/guides/webhooks). ### Parsing webhook payloads For most use cases, you will likely want to verify the webhook and parse the payload at the same time. To achieve this, we provide the method `client.webhooks.unwrap()`, which parses a webhook request and verifies that it was sent by OpenAI. This method will raise an error if the signature is invalid. Note that the `body` parameter must be the raw JSON string sent from the server (do not parse it first). The `.unwrap()` method will parse this JSON for you into an event object after verifying the webhook was sent from OpenAI. ``` from openai import OpenAI from flask import Flask, request app = Flask(__name__) client = OpenAI() # OPENAI_WEBHOOK_SECRET environment variable is used by default @app.route("/webhook", methods=["POST"]) def webhook(): request_body = request.get_data(as_text=True) try: event = client.webhooks.unwrap(request_body, request.headers) if event.type == "response.completed": print("Response completed:", event.data) elif event.type == "response.failed": print("Response failed:", event.data) else: print("Unhandled event type:", event.type) return "ok" except Exception as e: print("Invalid signature:", e) return "Invalid signature", 400 if __name__ == "__main__": app.run(port=8000) ``` ### Verifying webhook payloads directly In some cases, you may want to verify the webhook separately from parsing the payload. If you prefer to handle these steps separately, we provide the method `client.webhooks.verify_signature()` to *only verify* the signature of a webhook request. Like `.unwrap()`, this method will raise an error if the signature is invalid. Note that the `body` parameter must be the raw JSON string sent from the server (do not parse it first). You will then need to parse the body after verifying the signature. ``` import json from openai import OpenAI from flask import Flask, request app = Flask(__name__) client = OpenAI() # OPENAI_WEBHOOK_SECRET environment variable is used by default @app.route("/webhook", methods=["POST"]) def webhook(): request_body = request.get_data(as_text=True) try: client.webhooks.verify_signature(request_body, request.headers) # Parse the body after verification event = json.loads(request_body) print("Verified event:", event) return "ok" except Exception as e: print("Invalid signature:", e) return "Invalid signature", 400 if __name__ == "__main__": app.run(port=8000) ``` ## Handling errors When the library is unable to connect to the API (for example, due to network connection problems or a timeout), a subclass of `openai.APIConnectionError` is raised. When the API returns a non-success status code (that is, 4xx or 5xx response), a subclass of `openai.APIStatusError` is raised, containing `status_code` and `response` properties. All errors inherit from `openai.APIError`. ``` import openai from openai import OpenAI client = OpenAI() try: client.fine_tuning.jobs.create( model="gpt-4o", training_file="file-abc123", ) except openai.APIConnectionError as e: print("The server could not be reached") print(e.__cause__) # an underlying Exception, likely raised within httpx. except openai.RateLimitError as e: print("A 429 status code was received; we should back off a bit.") except openai.APIStatusError as e: print("Another non-200-range status code was received") print(e.status_code) print(e.response) ``` Error codes are as follows: | Status Code | Error Type | |---|---| | 400 | `BadRequestError` | | 401 | `AuthenticationError` | | 403 | `PermissionDeniedError` | | 404 | `NotFoundError` | | 422 | `UnprocessableEntityError` | | 429 | `RateLimitError` | | \>=500 | `InternalServerError` | | N/A | `APIConnectionError` | ## Request IDs > For more information on debugging requests, see [these docs](https://platform.openai.com/docs/api-reference/debugging-requests) All object responses in the SDK provide a `_request_id` property which is added from the `x-request-id` response header so that you can quickly log failing requests and report them back to OpenAI. ``` response = await client.responses.create( model="gpt-5.2", input="Say 'this is a test'.", ) print(response._request_id) # req_123 ``` Note that unlike other properties that use an `_` prefix, the `_request_id` property *is* public. Unless documented otherwise, *all* other `_` prefix properties, methods and modules are *private*. > \[!IMPORTANT\] > If you need to access request IDs for failed requests you must catch the `APIStatusError` exception ``` import openai try: completion = await client.chat.completions.create( messages=[{"role": "user", "content": "Say this is a test"}], model="gpt-5.2" ) except openai.APIStatusError as exc: print(exc.request_id) # req_123 raise exc ``` ## Retries Certain errors are automatically retried 2 times by default, with a short exponential backoff. Connection errors (for example, due to a network connectivity problem), 408 Request Timeout, 409 Conflict, 429 Rate Limit, and \>=500 Internal errors are all retried by default. You can use the `max_retries` option to configure or disable retry settings: ``` from openai import OpenAI # Configure the default for all requests: client = OpenAI( # default is 2 max_retries=0, ) # Or, configure per-request: client.with_options(max_retries=5).chat.completions.create( messages=[ { "role": "user", "content": "How can I get the name of the current day in JavaScript?", } ], model="gpt-5.2", ) ``` ## Timeouts By default requests time out after 10 minutes. You can configure this with a `timeout` option, which accepts a float or an [`httpx.Timeout`](https://www.python-httpx.org/advanced/timeouts/#fine-tuning-the-configuration) object: ``` from openai import OpenAI # Configure the default for all requests: client = OpenAI( # 20 seconds (default is 10 minutes) timeout=20.0, ) # More granular control: client = OpenAI( timeout=httpx.Timeout(60.0, read=5.0, write=10.0, connect=2.0), ) # Override per-request: client.with_options(timeout=5.0).chat.completions.create( messages=[ { "role": "user", "content": "How can I list all files in a directory using Python?", } ], model="gpt-5.2", ) ``` On timeout, an `APITimeoutError` is thrown. Note that requests that time out are [retried twice by default](https://github.com/openai/openai-python/tree/main/#retries). ## Advanced ### Logging We use the standard library [`logging`](https://docs.python.org/3/library/logging.html) module. You can enable logging by setting the environment variable `OPENAI_LOG` to `info`. ``` $ export OPENAI_LOG=info ``` Or to `debug` for more verbose logging. ### How to tell whether `None` means `null` or missing In an API response, a field may be explicitly `null`, or missing entirely; in either case, its value is `None` in this library. You can differentiate the two cases with `.model_fields_set`: ``` if response.my_field is None: if 'my_field' not in response.model_fields_set: print('Got json like {}, without a "my_field" key present at all.') else: print('Got json like {"my_field": null}.') ``` ### Accessing raw response data (e.g. headers) The "raw" Response object can be accessed by prefixing `.with_raw_response.` to any HTTP method call, e.g., ``` from openai import OpenAI client = OpenAI() response = client.chat.completions.with_raw_response.create( messages=[{ "role": "user", "content": "Say this is a test", }], model="gpt-5.2", ) print(response.headers.get('X-My-Header')) completion = response.parse() # get the object that `chat.completions.create()` would have returned print(completion) ``` These methods return a [`LegacyAPIResponse`](https://github.com/openai/openai-python/tree/main/src/openai/_legacy_response.py) object. This is a legacy class as we're changing it slightly in the next major version. For the sync client this will mostly be the same with the exception of `content` & `text` will be methods instead of properties. In the async client, all methods will be async. A migration script will be provided & the migration in general should be smooth. #### `.with_streaming_response` The above interface eagerly reads the full response body when you make the request, which may not always be what you want. To stream the response body, use `.with_streaming_response` instead, which requires a context manager and only reads the response body once you call `.read()`, `.text()`, `.json()`, `.iter_bytes()`, `.iter_text()`, `.iter_lines()` or `.parse()`. In the async client, these are async methods. As such, `.with_streaming_response` methods return a different [`APIResponse`](https://github.com/openai/openai-python/tree/main/src/openai/_response.py) object, and the async client returns an [`AsyncAPIResponse`](https://github.com/openai/openai-python/tree/main/src/openai/_response.py) object. ``` with client.chat.completions.with_streaming_response.create( messages=[ { "role": "user", "content": "Say this is a test", } ], model="gpt-5.2", ) as response: print(response.headers.get("X-My-Header")) for line in response.iter_lines(): print(line) ``` The context manager is required so that the response will reliably be closed. ### Making custom/undocumented requests This library is typed for convenient access to the documented API. If you need to access undocumented endpoints, params, or response properties, the library can still be used. #### Undocumented endpoints To make requests to undocumented endpoints, you can make requests using `client.get`, `client.post`, and other http verbs. Options on the client will be respected (such as retries) when making this request. ``` import httpx response = client.post( "/foo", cast_to=httpx.Response, body={"my_param": True}, ) print(response.headers.get("x-foo")) ``` #### Undocumented request params If you want to explicitly send an extra param, you can do so with the `extra_query`, `extra_body`, and `extra_headers` request options. #### Undocumented response properties To access undocumented response properties, you can access the extra fields like `response.unknown_prop`. You can also get all the extra fields on the Pydantic model as a dict with [`response.model_extra`](https://docs.pydantic.dev/latest/api/base_model/#pydantic.BaseModel.model_extra). ### Configuring the HTTP client You can directly override the [httpx client](https://www.python-httpx.org/api/#client) to customize it for your use case, including: - Support for [proxies](https://www.python-httpx.org/advanced/proxies/) - Custom [transports](https://www.python-httpx.org/advanced/transports/) - Additional [advanced](https://www.python-httpx.org/advanced/clients/) functionality ``` import httpx from openai import OpenAI, DefaultHttpxClient client = OpenAI( # Or use the `OPENAI_BASE_URL` env var base_url="http://my.test.server.example.com:8083/v1", http_client=DefaultHttpxClient( proxy="http://my.test.proxy.example.com", transport=httpx.HTTPTransport(local_address="0.0.0.0"), ), ) ``` You can also customize the client on a per-request basis by using `with_options()`: ``` client.with_options(http_client=DefaultHttpxClient(...)) ``` ### Managing HTTP resources By default the library closes underlying HTTP connections whenever the client is [garbage collected](https://docs.python.org/3/reference/datamodel.html#object.__del__). You can manually close the client using the `.close()` method if desired, or with a context manager that closes when exiting. ``` from openai import OpenAI with OpenAI() as client: # make requests here ... # HTTP client is now closed ``` ## Microsoft Azure OpenAI To use this library with [Azure OpenAI](https://learn.microsoft.com/azure/ai-services/openai/overview), use the `AzureOpenAI` class instead of the `OpenAI` class. > \[!IMPORTANT\] The Azure API shape differs from the core API shape which means that the static types for responses / params won't always be correct. ``` from openai import AzureOpenAI # gets the API Key from environment variable AZURE_OPENAI_API_KEY client = AzureOpenAI( # https://learn.microsoft.com/azure/ai-services/openai/reference#rest-api-versioning api_version="2023-07-01-preview", # https://learn.microsoft.com/azure/cognitive-services/openai/how-to/create-resource?pivots=web-portal#create-a-resource azure_endpoint="https://example-endpoint.openai.azure.com", ) completion = client.chat.completions.create( model="deployment-name", # e.g. gpt-35-instant messages=[ { "role": "user", "content": "How do I output all files in a directory using Python?", }, ], ) print(completion.to_json()) ``` In addition to the options provided in the base `OpenAI` client, the following options are provided: - `azure_endpoint` (or the `AZURE_OPENAI_ENDPOINT` environment variable) - `azure_deployment` - `api_version` (or the `OPENAI_API_VERSION` environment variable) - `azure_ad_token` (or the `AZURE_OPENAI_AD_TOKEN` environment variable) - `azure_ad_token_provider` An example of using the client with Microsoft Entra ID (formerly known as Azure Active Directory) can be found [here](https://github.com/openai/openai-python/blob/main/examples/azure_ad.py). ## Versioning This package generally follows [SemVer](https://semver.org/spec/v2.0.0.html) conventions, though certain backwards-incompatible changes may be released as minor versions: 1. Changes that only affect static types, without breaking runtime behavior. 2. Changes to library internals which are technically public but not intended or documented for external use. *(Please open a GitHub issue to let us know if you are relying on such internals.)* 3. Changes that we do not expect to impact the vast majority of users in practice. We take backwards-compatibility seriously and work hard to ensure you can rely on a smooth upgrade experience. We are keen for your feedback; please open an [issue](https://www.github.com/openai/openai-python/issues) with questions, bugs, or suggestions. ### Determining the installed version If you've upgraded to the latest version but aren't seeing any new features you were expecting then your python environment is likely still using an older version. You can determine the version that is being used at runtime with: ``` import openai print(openai.__version__) ``` ## Requirements Python 3.9 or higher. ## Contributing See [the contributing documentation](https://github.com/openai/openai-python/tree/main/CONTRIBUTING.md). ## Project details ### Verified details *These details have been [verified by PyPI](https://docs.pypi.org/project_metadata/#verified-details)* ###### Owner - [OpenAI](https://pypi.org/org/openai/) ###### Maintainers [![Avatar for ddeville from gravatar.com](https://pypi-camo.freetls.fastly.net/32f89b8b480b8a373a82e43b9846aa7f3d1347e3/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f36356430636636356336623363303534396439316630303331333932626234343f73697a653d3530) ddeville](https://pypi.org/user/ddeville/) [![Avatar for dschnurr-openai from gravatar.com](https://pypi-camo.freetls.fastly.net/d276ae7e441dadd38c03dc2cddfc79a40bdfe6f4/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f39396434623863633962623533346639653134383466393432316261343765363f73697a653d3530) dschnurr-openai](https://pypi.org/user/dschnurr-openai/) [![Avatar for gdb from gravatar.com](https://pypi-camo.freetls.fastly.net/714c5137f119ec3e27cb0fc83a0042474855d1ff/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f39643862336538633662616266666461336366353063633166663863383731653f73697a653d3530) gdb](https://pypi.org/user/gdb/) [![Avatar for hponde\_oai from gravatar.com](https://pypi-camo.freetls.fastly.net/41ade05ccfd7c00016ca9067cfa93c96fae5c47c/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f66313362623930363033383863393761383663656261323239646534383633613f73697a653d3530) hponde\_oai](https://pypi.org/user/hponde_oai/) [![Avatar for jhallard from gravatar.com](https://pypi-camo.freetls.fastly.net/3eda32bb127f92256b2848a48a6a42a085858f86/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f61343634643164613232313863663031343866323334303230353766356333323f73697a653d3530) jhallard](https://pypi.org/user/jhallard/) [![Avatar for tomerkOpenAI from gravatar.com](https://pypi-camo.freetls.fastly.net/2eeda1a7257b4a01b3a5b7c9c3f21298ab30711b/68747470733a2f2f7365637572652e67726176617461722e636f6d2f6176617461722f35396231346638626664326633303562343632373563346231373466663165643f73697a653d3530) tomerkOpenAI](https://pypi.org/user/tomerkOpenAI/) ### Unverified details *These details have **not** been verified by PyPI* ###### Project links - [Homepage](https://github.com/openai/openai-python) - [Repository](https://github.com/openai/openai-python) ###### Meta - **License:** Apache Software License (Apache-2.0) - **Author:** [OpenAI](mailto:support@openai.com) - **Requires:** Python \>=3.9 - **Provides-Extra:** `aiohttp` , `datalib` , `realtime` , `voice-helpers` ###### Classifiers - **Intended Audience** - [Developers](https://pypi.org/search/?c=Intended+Audience+%3A%3A+Developers) - **License** - [OSI Approved :: Apache Software License](https://pypi.org/search/?c=License+%3A%3A+OSI+Approved+%3A%3A+Apache+Software+License) - **Operating System** - [MacOS](https://pypi.org/search/?c=Operating+System+%3A%3A+MacOS) - [Microsoft :: Windows](https://pypi.org/search/?c=Operating+System+%3A%3A+Microsoft+%3A%3A+Windows) - [OS Independent](https://pypi.org/search/?c=Operating+System+%3A%3A+OS+Independent) - [POSIX](https://pypi.org/search/?c=Operating+System+%3A%3A+POSIX) - [POSIX :: Linux](https://pypi.org/search/?c=Operating+System+%3A%3A+POSIX+%3A%3A+Linux) - **Programming Language** - [Python :: 3.9](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.9) - [Python :: 3.10](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.10) - [Python :: 3.11](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.11) - [Python :: 3.12](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.12) - [Python :: 3.13](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.13) - [Python :: 3.14](https://pypi.org/search/?c=Programming+Language+%3A%3A+Python+%3A%3A+3.14) - **Topic** - [Software Development :: Libraries :: Python Modules](https://pypi.org/search/?c=Topic+%3A%3A+Software+Development+%3A%3A+Libraries+%3A%3A+Python+Modules) - **Typing** - [Typed](https://pypi.org/search/?c=Typing+%3A%3A+Typed) ## Release history [Release notifications](https://pypi.org/help/#project-release-notifications) \| [RSS feed](https://pypi.org/rss/project/openai/releases.xml) This version ![](https://pypi.org/static/images/blue-cube.572a5bfb.svg) [2\.31.0 Apr 8, 2026](https://pypi.org/project/openai/2.31.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.30.0 Mar 25, 2026](https://pypi.org/project/openai/2.30.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.29.0 Mar 17, 2026](https://pypi.org/project/openai/2.29.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.28.0 Mar 13, 2026](https://pypi.org/project/openai/2.28.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.27.0 Mar 13, 2026](https://pypi.org/project/openai/2.27.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.26.0 Mar 5, 2026](https://pypi.org/project/openai/2.26.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.25.0 Mar 5, 2026](https://pypi.org/project/openai/2.25.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.24.0 Feb 24, 2026](https://pypi.org/project/openai/2.24.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.23.0 Feb 24, 2026](https://pypi.org/project/openai/2.23.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.22.0 Feb 23, 2026](https://pypi.org/project/openai/2.22.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.21.0 Feb 14, 2026](https://pypi.org/project/openai/2.21.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.20.0 Feb 10, 2026](https://pypi.org/project/openai/2.20.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.19.0 Feb 10, 2026](https://pypi.org/project/openai/2.19.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.18.0 Feb 9, 2026](https://pypi.org/project/openai/2.18.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.17.0 Feb 5, 2026](https://pypi.org/project/openai/2.17.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.16.0 Jan 27, 2026](https://pypi.org/project/openai/2.16.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.15.0 Jan 9, 2026](https://pypi.org/project/openai/2.15.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.14.0 Dec 19, 2025](https://pypi.org/project/openai/2.14.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.13.0 Dec 16, 2025](https://pypi.org/project/openai/2.13.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.12.0 Dec 15, 2025](https://pypi.org/project/openai/2.12.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.11.0 Dec 11, 2025](https://pypi.org/project/openai/2.11.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.9.0 Dec 4, 2025](https://pypi.org/project/openai/2.9.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.8.1 Nov 17, 2025](https://pypi.org/project/openai/2.8.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.8.0 Nov 13, 2025](https://pypi.org/project/openai/2.8.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.7.2 Nov 10, 2025](https://pypi.org/project/openai/2.7.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.7.1 Nov 4, 2025](https://pypi.org/project/openai/2.7.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.7.0 Nov 3, 2025](https://pypi.org/project/openai/2.7.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.6.1 Oct 24, 2025](https://pypi.org/project/openai/2.6.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.6.0 Oct 20, 2025](https://pypi.org/project/openai/2.6.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.5.0 Oct 17, 2025](https://pypi.org/project/openai/2.5.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.4.0 Oct 16, 2025](https://pypi.org/project/openai/2.4.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.3.0 Oct 10, 2025](https://pypi.org/project/openai/2.3.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.2.0 Oct 6, 2025](https://pypi.org/project/openai/2.2.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.1.0 Oct 2, 2025](https://pypi.org/project/openai/2.1.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.0.1 Oct 1, 2025](https://pypi.org/project/openai/2.0.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [2\.0.0 Sep 30, 2025](https://pypi.org/project/openai/2.0.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.109.1 Sep 24, 2025](https://pypi.org/project/openai/1.109.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.109.0 Sep 23, 2025](https://pypi.org/project/openai/1.109.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.108.2 Sep 22, 2025](https://pypi.org/project/openai/1.108.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.108.1 Sep 19, 2025](https://pypi.org/project/openai/1.108.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.108.0 Sep 17, 2025](https://pypi.org/project/openai/1.108.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.107.3 Sep 15, 2025](https://pypi.org/project/openai/1.107.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.107.2 Sep 12, 2025](https://pypi.org/project/openai/1.107.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.107.1 Sep 10, 2025](https://pypi.org/project/openai/1.107.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.107.0 Sep 8, 2025](https://pypi.org/project/openai/1.107.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.106.1 Sep 4, 2025](https://pypi.org/project/openai/1.106.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.106.0 Sep 4, 2025](https://pypi.org/project/openai/1.106.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.105.0 Sep 3, 2025](https://pypi.org/project/openai/1.105.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.104.2 Sep 2, 2025](https://pypi.org/project/openai/1.104.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.104.1 Sep 2, 2025](https://pypi.org/project/openai/1.104.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.104.0 Sep 2, 2025](https://pypi.org/project/openai/1.104.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.103.0 Sep 2, 2025](https://pypi.org/project/openai/1.103.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.102.0 Aug 26, 2025](https://pypi.org/project/openai/1.102.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.101.0 Aug 21, 2025](https://pypi.org/project/openai/1.101.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.100.2 Aug 19, 2025](https://pypi.org/project/openai/1.100.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.100.1 Aug 18, 2025](https://pypi.org/project/openai/1.100.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.100.0 Aug 18, 2025](https://pypi.org/project/openai/1.100.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.9 Aug 12, 2025](https://pypi.org/project/openai/1.99.9/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.8 Aug 11, 2025](https://pypi.org/project/openai/1.99.8/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.7 Aug 11, 2025](https://pypi.org/project/openai/1.99.7/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.6 Aug 9, 2025](https://pypi.org/project/openai/1.99.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.5 Aug 8, 2025](https://pypi.org/project/openai/1.99.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.4 Aug 8, 2025](https://pypi.org/project/openai/1.99.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.3 Aug 7, 2025](https://pypi.org/project/openai/1.99.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.2 Aug 7, 2025](https://pypi.org/project/openai/1.99.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.1 Aug 5, 2025](https://pypi.org/project/openai/1.99.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.99.0 Aug 5, 2025](https://pypi.org/project/openai/1.99.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.98.0 Jul 30, 2025](https://pypi.org/project/openai/1.98.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.97.2 Jul 30, 2025](https://pypi.org/project/openai/1.97.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.97.1 Jul 22, 2025](https://pypi.org/project/openai/1.97.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.97.0 Jul 16, 2025](https://pypi.org/project/openai/1.97.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.96.1 Jul 15, 2025](https://pypi.org/project/openai/1.96.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.96.0 Jul 15, 2025](https://pypi.org/project/openai/1.96.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.95.1 Jul 11, 2025](https://pypi.org/project/openai/1.95.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.95.0 Jul 10, 2025](https://pypi.org/project/openai/1.95.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.94.0 Jul 10, 2025](https://pypi.org/project/openai/1.94.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.93.3 Jul 9, 2025](https://pypi.org/project/openai/1.93.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.93.2 Jul 8, 2025](https://pypi.org/project/openai/1.93.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.93.1 Jul 7, 2025](https://pypi.org/project/openai/1.93.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.93.0 Jun 27, 2025](https://pypi.org/project/openai/1.93.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.92.3 Jun 27, 2025](https://pypi.org/project/openai/1.92.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.92.2 Jun 26, 2025](https://pypi.org/project/openai/1.92.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.92.1 Jun 26, 2025](https://pypi.org/project/openai/1.92.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.92.0 Jun 26, 2025](https://pypi.org/project/openai/1.92.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.91.0 Jun 23, 2025](https://pypi.org/project/openai/1.91.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.90.0 Jun 20, 2025](https://pypi.org/project/openai/1.90.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.89.0 Jun 20, 2025](https://pypi.org/project/openai/1.89.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.88.0 Jun 17, 2025](https://pypi.org/project/openai/1.88.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.87.0 Jun 16, 2025](https://pypi.org/project/openai/1.87.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.86.0 Jun 10, 2025](https://pypi.org/project/openai/1.86.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.85.0 Jun 9, 2025](https://pypi.org/project/openai/1.85.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.84.0 Jun 3, 2025](https://pypi.org/project/openai/1.84.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.83.0 Jun 2, 2025](https://pypi.org/project/openai/1.83.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.82.1 May 29, 2025](https://pypi.org/project/openai/1.82.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.82.0 May 22, 2025](https://pypi.org/project/openai/1.82.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.81.0 May 21, 2025](https://pypi.org/project/openai/1.81.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.80.0 May 21, 2025](https://pypi.org/project/openai/1.80.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.79.0 May 16, 2025](https://pypi.org/project/openai/1.79.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.78.1 May 12, 2025](https://pypi.org/project/openai/1.78.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.78.0 May 8, 2025](https://pypi.org/project/openai/1.78.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.77.0 May 2, 2025](https://pypi.org/project/openai/1.77.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.76.2 Apr 29, 2025](https://pypi.org/project/openai/1.76.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.76.1 Apr 29, 2025](https://pypi.org/project/openai/1.76.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.76.0 Apr 23, 2025](https://pypi.org/project/openai/1.76.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.75.0 Apr 16, 2025](https://pypi.org/project/openai/1.75.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.74.1 Apr 16, 2025](https://pypi.org/project/openai/1.74.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.74.0 Apr 14, 2025](https://pypi.org/project/openai/1.74.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.73.0 Apr 12, 2025](https://pypi.org/project/openai/1.73.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.72.0 Apr 8, 2025](https://pypi.org/project/openai/1.72.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.71.0 Apr 7, 2025](https://pypi.org/project/openai/1.71.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.70.0 Mar 31, 2025](https://pypi.org/project/openai/1.70.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.69.0 Mar 27, 2025](https://pypi.org/project/openai/1.69.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.68.2 Mar 21, 2025](https://pypi.org/project/openai/1.68.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.68.1 Mar 21, 2025](https://pypi.org/project/openai/1.68.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.68.0 Mar 20, 2025](https://pypi.org/project/openai/1.68.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.67.0 Mar 19, 2025](https://pypi.org/project/openai/1.67.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.66.5 Mar 18, 2025](https://pypi.org/project/openai/1.66.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.66.3 Mar 12, 2025](https://pypi.org/project/openai/1.66.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.66.2 Mar 11, 2025](https://pypi.org/project/openai/1.66.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.66.1 Mar 11, 2025](https://pypi.org/project/openai/1.66.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.66.0 Mar 11, 2025](https://pypi.org/project/openai/1.66.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.65.5 Mar 9, 2025](https://pypi.org/project/openai/1.65.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.65.4 Mar 5, 2025](https://pypi.org/project/openai/1.65.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.65.3 Mar 4, 2025](https://pypi.org/project/openai/1.65.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.65.2 Mar 1, 2025](https://pypi.org/project/openai/1.65.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.65.1 Feb 27, 2025](https://pypi.org/project/openai/1.65.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.65.0 Feb 27, 2025](https://pypi.org/project/openai/1.65.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.64.0 Feb 22, 2025](https://pypi.org/project/openai/1.64.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.63.2 Feb 17, 2025](https://pypi.org/project/openai/1.63.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.63.1 Feb 17, 2025](https://pypi.org/project/openai/1.63.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.63.0 Feb 13, 2025](https://pypi.org/project/openai/1.63.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.62.0 Feb 12, 2025](https://pypi.org/project/openai/1.62.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.61.1 Feb 5, 2025](https://pypi.org/project/openai/1.61.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.61.0 Jan 31, 2025](https://pypi.org/project/openai/1.61.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.60.2 Jan 27, 2025](https://pypi.org/project/openai/1.60.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.60.1 Jan 24, 2025](https://pypi.org/project/openai/1.60.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.60.0 Jan 22, 2025](https://pypi.org/project/openai/1.60.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.9 Jan 20, 2025](https://pypi.org/project/openai/1.59.9/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.8 Jan 17, 2025](https://pypi.org/project/openai/1.59.8/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.7 Jan 13, 2025](https://pypi.org/project/openai/1.59.7/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.6 Jan 9, 2025](https://pypi.org/project/openai/1.59.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.5 Jan 8, 2025](https://pypi.org/project/openai/1.59.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.4 Jan 7, 2025](https://pypi.org/project/openai/1.59.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.3 Jan 3, 2025](https://pypi.org/project/openai/1.59.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.59.2 Jan 3, 2025](https://pypi.org/project/openai/1.59.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.58.1 Dec 17, 2024](https://pypi.org/project/openai/1.58.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.58.0 Dec 17, 2024](https://pypi.org/project/openai/1.58.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.57.4 Dec 13, 2024](https://pypi.org/project/openai/1.57.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.57.3 Dec 12, 2024](https://pypi.org/project/openai/1.57.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.57.2 Dec 10, 2024](https://pypi.org/project/openai/1.57.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.57.1 Dec 9, 2024](https://pypi.org/project/openai/1.57.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.57.0 Dec 5, 2024](https://pypi.org/project/openai/1.57.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.56.2 Dec 4, 2024](https://pypi.org/project/openai/1.56.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.56.1 Dec 3, 2024](https://pypi.org/project/openai/1.56.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.56.0 Dec 2, 2024](https://pypi.org/project/openai/1.56.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.55.3 Nov 28, 2024](https://pypi.org/project/openai/1.55.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.55.2 Nov 27, 2024](https://pypi.org/project/openai/1.55.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.55.1 Nov 25, 2024](https://pypi.org/project/openai/1.55.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.55.0 Nov 20, 2024](https://pypi.org/project/openai/1.55.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.54.5 Nov 19, 2024](https://pypi.org/project/openai/1.54.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.54.4 Nov 12, 2024](https://pypi.org/project/openai/1.54.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.54.3 Nov 6, 2024](https://pypi.org/project/openai/1.54.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.54.2 Nov 6, 2024](https://pypi.org/project/openai/1.54.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.54.1 Nov 5, 2024](https://pypi.org/project/openai/1.54.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.54.0 Nov 4, 2024](https://pypi.org/project/openai/1.54.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.53.1 Nov 4, 2024](https://pypi.org/project/openai/1.53.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.53.0 Oct 30, 2024](https://pypi.org/project/openai/1.53.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.52.2 Oct 23, 2024](https://pypi.org/project/openai/1.52.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.52.1 Oct 22, 2024](https://pypi.org/project/openai/1.52.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.52.0 Oct 17, 2024](https://pypi.org/project/openai/1.52.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.51.2 Oct 8, 2024](https://pypi.org/project/openai/1.51.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.51.1 Oct 7, 2024](https://pypi.org/project/openai/1.51.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.51.0 Oct 1, 2024](https://pypi.org/project/openai/1.51.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.50.2 Sep 27, 2024](https://pypi.org/project/openai/1.50.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.50.1 Sep 27, 2024](https://pypi.org/project/openai/1.50.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.50.0 Sep 26, 2024](https://pypi.org/project/openai/1.50.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.49.0 Sep 26, 2024](https://pypi.org/project/openai/1.49.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.48.0 Sep 25, 2024](https://pypi.org/project/openai/1.48.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.47.1 Sep 23, 2024](https://pypi.org/project/openai/1.47.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.47.0 Sep 20, 2024](https://pypi.org/project/openai/1.47.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.46.1 Sep 19, 2024](https://pypi.org/project/openai/1.46.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.46.0 Sep 17, 2024](https://pypi.org/project/openai/1.46.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.45.1 Sep 16, 2024](https://pypi.org/project/openai/1.45.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.45.0 Sep 12, 2024](https://pypi.org/project/openai/1.45.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.44.1 Sep 9, 2024](https://pypi.org/project/openai/1.44.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.44.0 Sep 6, 2024](https://pypi.org/project/openai/1.44.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.43.1 Sep 5, 2024](https://pypi.org/project/openai/1.43.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.43.0 Aug 29, 2024](https://pypi.org/project/openai/1.43.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.42.0 Aug 20, 2024](https://pypi.org/project/openai/1.42.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.41.1 Aug 20, 2024](https://pypi.org/project/openai/1.41.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.41.0 Aug 16, 2024](https://pypi.org/project/openai/1.41.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.8 Aug 15, 2024](https://pypi.org/project/openai/1.40.8/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.7 Aug 15, 2024](https://pypi.org/project/openai/1.40.7/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.6 Aug 12, 2024](https://pypi.org/project/openai/1.40.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.5 Aug 12, 2024](https://pypi.org/project/openai/1.40.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.4 Aug 12, 2024](https://pypi.org/project/openai/1.40.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.3 Aug 10, 2024](https://pypi.org/project/openai/1.40.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.2 Aug 8, 2024](https://pypi.org/project/openai/1.40.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.1 Aug 7, 2024](https://pypi.org/project/openai/1.40.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.40.0 Aug 6, 2024](https://pypi.org/project/openai/1.40.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.39.0 Aug 5, 2024](https://pypi.org/project/openai/1.39.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.38.0 Aug 2, 2024](https://pypi.org/project/openai/1.38.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.37.2 Aug 2, 2024](https://pypi.org/project/openai/1.37.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.37.1 Jul 25, 2024](https://pypi.org/project/openai/1.37.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.37.0 Jul 22, 2024](https://pypi.org/project/openai/1.37.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.36.1 Jul 20, 2024](https://pypi.org/project/openai/1.36.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.36.0 Jul 19, 2024](https://pypi.org/project/openai/1.36.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.15 Jul 18, 2024](https://pypi.org/project/openai/1.35.15/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.14 Jul 16, 2024](https://pypi.org/project/openai/1.35.14/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.13 Jul 10, 2024](https://pypi.org/project/openai/1.35.13/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.12 Jul 9, 2024](https://pypi.org/project/openai/1.35.12/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.11 Jul 9, 2024](https://pypi.org/project/openai/1.35.11/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.10 Jul 4, 2024](https://pypi.org/project/openai/1.35.10/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.9 Jul 2, 2024](https://pypi.org/project/openai/1.35.9/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.8 Jul 2, 2024](https://pypi.org/project/openai/1.35.8/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.7 Jun 27, 2024](https://pypi.org/project/openai/1.35.7/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.6 Jun 27, 2024](https://pypi.org/project/openai/1.35.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.5 Jun 26, 2024](https://pypi.org/project/openai/1.35.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.4 Jun 26, 2024](https://pypi.org/project/openai/1.35.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.3 Jun 20, 2024](https://pypi.org/project/openai/1.35.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.2 Jun 20, 2024](https://pypi.org/project/openai/1.35.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.1 Jun 19, 2024](https://pypi.org/project/openai/1.35.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.35.0 Jun 19, 2024](https://pypi.org/project/openai/1.35.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.34.0 Jun 12, 2024](https://pypi.org/project/openai/1.34.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.33.0 Jun 7, 2024](https://pypi.org/project/openai/1.33.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.32.1 Jun 7, 2024](https://pypi.org/project/openai/1.32.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.32.0 Jun 6, 2024](https://pypi.org/project/openai/1.32.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.31.2 Jun 6, 2024](https://pypi.org/project/openai/1.31.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.31.1 Jun 5, 2024](https://pypi.org/project/openai/1.31.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.31.0 Jun 3, 2024](https://pypi.org/project/openai/1.31.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.30.5 May 30, 2024](https://pypi.org/project/openai/1.30.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.30.4 May 28, 2024](https://pypi.org/project/openai/1.30.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.30.3 May 24, 2024](https://pypi.org/project/openai/1.30.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.30.2 May 23, 2024](https://pypi.org/project/openai/1.30.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.30.1 May 14, 2024](https://pypi.org/project/openai/1.30.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.30.0 May 14, 2024](https://pypi.org/project/openai/1.30.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.29.0 May 13, 2024](https://pypi.org/project/openai/1.29.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.28.2 May 13, 2024](https://pypi.org/project/openai/1.28.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.28.1 May 11, 2024](https://pypi.org/project/openai/1.28.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.28.0 May 9, 2024](https://pypi.org/project/openai/1.28.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.27.0 May 8, 2024](https://pypi.org/project/openai/1.27.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.26.0 May 6, 2024](https://pypi.org/project/openai/1.26.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.25.2 May 5, 2024](https://pypi.org/project/openai/1.25.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.25.1 May 2, 2024](https://pypi.org/project/openai/1.25.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.25.0 May 1, 2024](https://pypi.org/project/openai/1.25.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.24.1 Apr 30, 2024](https://pypi.org/project/openai/1.24.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.24.0 Apr 29, 2024](https://pypi.org/project/openai/1.24.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.23.6 Apr 25, 2024](https://pypi.org/project/openai/1.23.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.23.5 Apr 25, 2024](https://pypi.org/project/openai/1.23.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.23.4 Apr 24, 2024](https://pypi.org/project/openai/1.23.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.23.3 Apr 23, 2024](https://pypi.org/project/openai/1.23.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.23.2 Apr 19, 2024](https://pypi.org/project/openai/1.23.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.23.1 Apr 18, 2024](https://pypi.org/project/openai/1.23.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.23.0 Apr 18, 2024](https://pypi.org/project/openai/1.23.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.22.0 Apr 18, 2024](https://pypi.org/project/openai/1.22.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.21.2 Apr 17, 2024](https://pypi.org/project/openai/1.21.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.21.1 Apr 17, 2024](https://pypi.org/project/openai/1.21.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.21.0 Apr 17, 2024](https://pypi.org/project/openai/1.21.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.20.0 Apr 16, 2024](https://pypi.org/project/openai/1.20.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.19.0 Apr 16, 2024](https://pypi.org/project/openai/1.19.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.18.0 Apr 15, 2024](https://pypi.org/project/openai/1.18.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.17.1 Apr 13, 2024](https://pypi.org/project/openai/1.17.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.17.0 Apr 10, 2024](https://pypi.org/project/openai/1.17.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.16.2 Apr 4, 2024](https://pypi.org/project/openai/1.16.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.16.1 Apr 2, 2024](https://pypi.org/project/openai/1.16.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.16.0 Apr 1, 2024](https://pypi.org/project/openai/1.16.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.14.3 Mar 25, 2024](https://pypi.org/project/openai/1.14.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.14.2 Mar 19, 2024](https://pypi.org/project/openai/1.14.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.14.1 Mar 15, 2024](https://pypi.org/project/openai/1.14.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.14.0 Mar 13, 2024](https://pypi.org/project/openai/1.14.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.13.4 Mar 13, 2024](https://pypi.org/project/openai/1.13.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.13.3 Feb 28, 2024](https://pypi.org/project/openai/1.13.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.12.0 Feb 9, 2024](https://pypi.org/project/openai/1.12.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.11.1 Feb 4, 2024](https://pypi.org/project/openai/1.11.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.11.0 Feb 3, 2024](https://pypi.org/project/openai/1.11.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.10.0 Jan 25, 2024](https://pypi.org/project/openai/1.10.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.9.0 Jan 21, 2024](https://pypi.org/project/openai/1.9.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.8.0 Jan 16, 2024](https://pypi.org/project/openai/1.8.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.7.2 Jan 12, 2024](https://pypi.org/project/openai/1.7.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.7.1 Jan 10, 2024](https://pypi.org/project/openai/1.7.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.7.0 Jan 9, 2024](https://pypi.org/project/openai/1.7.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.6.1 Dec 22, 2023](https://pypi.org/project/openai/1.6.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.6.0 Dec 19, 2023](https://pypi.org/project/openai/1.6.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.5.0 Dec 17, 2023](https://pypi.org/project/openai/1.5.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.4.0 Dec 15, 2023](https://pypi.org/project/openai/1.4.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.9 Dec 13, 2023](https://pypi.org/project/openai/1.3.9/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.8 Dec 9, 2023](https://pypi.org/project/openai/1.3.8/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.7 Dec 1, 2023](https://pypi.org/project/openai/1.3.7/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.6 Nov 29, 2023](https://pypi.org/project/openai/1.3.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.5 Nov 22, 2023](https://pypi.org/project/openai/1.3.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.4 Nov 21, 2023](https://pypi.org/project/openai/1.3.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.3 Nov 17, 2023](https://pypi.org/project/openai/1.3.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.2 Nov 16, 2023](https://pypi.org/project/openai/1.3.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.1 Nov 16, 2023](https://pypi.org/project/openai/1.3.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.3.0 Nov 15, 2023](https://pypi.org/project/openai/1.3.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.2.4 Nov 13, 2023](https://pypi.org/project/openai/1.2.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.2.3 Nov 10, 2023](https://pypi.org/project/openai/1.2.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.2.2 Nov 9, 2023](https://pypi.org/project/openai/1.2.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.2.1 Nov 9, 2023](https://pypi.org/project/openai/1.2.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.2.0 Nov 9, 2023](https://pypi.org/project/openai/1.2.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.1.2 Nov 8, 2023](https://pypi.org/project/openai/1.1.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.1.1 Nov 6, 2023](https://pypi.org/project/openai/1.1.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.1.0 Nov 6, 2023](https://pypi.org/project/openai/1.1.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.1 Nov 6, 2023](https://pypi.org/project/openai/1.0.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.0 Nov 6, 2023](https://pypi.org/project/openai/1.0.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.0rc3 pre-release Nov 6, 2023](https://pypi.org/project/openai/1.0.0rc3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.0rc2 pre-release Nov 3, 2023](https://pypi.org/project/openai/1.0.0rc2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.0rc1 pre-release Oct 28, 2023](https://pypi.org/project/openai/1.0.0rc1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.0b3 pre-release Oct 17, 2023](https://pypi.org/project/openai/1.0.0b3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.0b2 pre-release Oct 12, 2023](https://pypi.org/project/openai/1.0.0b2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [1\.0.0b1 pre-release Sep 29, 2023](https://pypi.org/project/openai/1.0.0b1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.28.1 Sep 26, 2023](https://pypi.org/project/openai/0.28.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.28.0 Aug 31, 2023](https://pypi.org/project/openai/0.28.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.10 Aug 30, 2023](https://pypi.org/project/openai/0.27.10/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.9 Aug 22, 2023](https://pypi.org/project/openai/0.27.9/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.8 Jun 7, 2023](https://pypi.org/project/openai/0.27.8/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.7 May 19, 2023](https://pypi.org/project/openai/0.27.7/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.6 May 1, 2023](https://pypi.org/project/openai/0.27.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.5 Apr 27, 2023](https://pypi.org/project/openai/0.27.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.4 Apr 4, 2023](https://pypi.org/project/openai/0.27.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.3 Apr 3, 2023](https://pypi.org/project/openai/0.27.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.2 Mar 11, 2023](https://pypi.org/project/openai/0.27.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.1 Mar 8, 2023](https://pypi.org/project/openai/0.27.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.27.0 Mar 1, 2023](https://pypi.org/project/openai/0.27.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.26.5 Feb 7, 2023](https://pypi.org/project/openai/0.26.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.26.4 Jan 26, 2023](https://pypi.org/project/openai/0.26.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.26.3 Jan 25, 2023](https://pypi.org/project/openai/0.26.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.26.2 Jan 24, 2023](https://pypi.org/project/openai/0.26.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.26.1 Jan 13, 2023](https://pypi.org/project/openai/0.26.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.26.0 Jan 6, 2023](https://pypi.org/project/openai/0.26.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.25.0 Nov 2, 2022](https://pypi.org/project/openai/0.25.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.24.0 Oct 21, 2022](https://pypi.org/project/openai/0.24.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.23.1 Sep 28, 2022](https://pypi.org/project/openai/0.23.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.23.0 Aug 24, 2022](https://pypi.org/project/openai/0.23.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.22.1 Aug 2, 2022](https://pypi.org/project/openai/0.22.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.22.0 Jul 26, 2022](https://pypi.org/project/openai/0.22.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.20.0 Jun 16, 2022](https://pypi.org/project/openai/0.20.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.19.0 May 24, 2022](https://pypi.org/project/openai/0.19.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.18.1 Apr 15, 2022](https://pypi.org/project/openai/0.18.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.18.0 Apr 8, 2022](https://pypi.org/project/openai/0.18.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.16.0 Mar 17, 2022](https://pypi.org/project/openai/0.16.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.15.0 Feb 24, 2022](https://pypi.org/project/openai/0.15.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.14.0 Feb 2, 2022](https://pypi.org/project/openai/0.14.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.13.0 Jan 25, 2022](https://pypi.org/project/openai/0.13.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.12.0 Jan 22, 2022](https://pypi.org/project/openai/0.12.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.11.6 Jan 21, 2022](https://pypi.org/project/openai/0.11.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.11.5 Dec 21, 2021](https://pypi.org/project/openai/0.11.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.11.4 Dec 14, 2021](https://pypi.org/project/openai/0.11.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.11.3 Dec 4, 2021](https://pypi.org/project/openai/0.11.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.11.2 Dec 4, 2021](https://pypi.org/project/openai/0.11.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.11.1 Dec 2, 2021](https://pypi.org/project/openai/0.11.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.11.0 Oct 27, 2021](https://pypi.org/project/openai/0.11.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.10.5 Oct 1, 2021](https://pypi.org/project/openai/0.10.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.10.4 Sep 9, 2021](https://pypi.org/project/openai/0.10.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.10.3 Aug 31, 2021](https://pypi.org/project/openai/0.10.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.10.2 Jul 29, 2021](https://pypi.org/project/openai/0.10.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.10.1 Jul 14, 2021](https://pypi.org/project/openai/0.10.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.10.0 Jul 14, 2021](https://pypi.org/project/openai/0.10.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.9.4 Jul 12, 2021](https://pypi.org/project/openai/0.9.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.9.3 Jun 30, 2021](https://pypi.org/project/openai/0.9.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.9.2 Jun 30, 2021](https://pypi.org/project/openai/0.9.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.9.1 Jun 30, 2021](https://pypi.org/project/openai/0.9.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.9.0 Jun 29, 2021](https://pypi.org/project/openai/0.9.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.8.0 Jun 17, 2021](https://pypi.org/project/openai/0.8.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.7.0 Jun 11, 2021](https://pypi.org/project/openai/0.7.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.6.4 May 21, 2021](https://pypi.org/project/openai/0.6.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.6.3 Apr 12, 2021](https://pypi.org/project/openai/0.6.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.6.2 Mar 20, 2021](https://pypi.org/project/openai/0.6.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.6.1 Mar 19, 2021](https://pypi.org/project/openai/0.6.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.6.0 Mar 18, 2021](https://pypi.org/project/openai/0.6.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.4.0 Mar 4, 2021](https://pypi.org/project/openai/0.4.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.3.0 Jan 27, 2021](https://pypi.org/project/openai/0.3.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.2.6 Oct 24, 2020](https://pypi.org/project/openai/0.2.6/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.2.5 Oct 5, 2020](https://pypi.org/project/openai/0.2.5/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.2.4 Jul 12, 2020](https://pypi.org/project/openai/0.2.4/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.2.3 Jul 7, 2020](https://pypi.org/project/openai/0.2.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.2.1 Jun 13, 2020](https://pypi.org/project/openai/0.2.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.2.0 Jun 13, 2020](https://pypi.org/project/openai/0.2.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.1.3 Jun 13, 2020](https://pypi.org/project/openai/0.1.3/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.1.2 Jun 13, 2020](https://pypi.org/project/openai/0.1.2/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.1.1 Jun 13, 2020](https://pypi.org/project/openai/0.1.1/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.1.0 Jun 13, 2020](https://pypi.org/project/openai/0.1.0/) ![](https://pypi.org/static/images/white-cube.2351a86c.svg) [0\.0.2 Feb 18, 2020](https://pypi.org/project/openai/0.0.2/) ## Download files Download the file for your platform. If you're not sure which to choose, learn more about [installing packages](https://packaging.python.org/tutorials/installing-packages/ "External link"). ### Source Distribution [openai-2.31.0.tar.gz](https://files.pythonhosted.org/packages/94/fe/64b3d035780b3188f86c4f6f1bc202e7bb74757ef028802112273b9dcacf/openai-2.31.0.tar.gz) (684.8 kB [view details](https://pypi.org/project/openai/#openai-2.31.0.tar.gz)) Uploaded Apr 8, 2026 `Source` ### Built Distribution Filter files by name, interpreter, ABI, and platform. If you're not sure about the file name format, learn more about [wheel file names](https://packaging.python.org/en/latest/specifications/binary-distribution-format/ "External link"). The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files. Copy a direct link to the current filters <https://pypi.org/project/openai/#files> Copy Showing 1 of 1 file. File name Interpreter ABI Platform [openai-2.31.0-py3-none-any.whl](https://files.pythonhosted.org/packages/66/bc/a8f7c3aa03452fedbb9af8be83e959adba96a6b4a35e416faffcc959c568/openai-2.31.0-py3-none-any.whl) (1.2 MB [view details](https://pypi.org/project/openai/#openai-2.31.0-py3-none-any.whl)) Uploaded Apr 8, 2026 `Python 3` ## File details Details for the file `openai-2.31.0.tar.gz`. ### File metadata - Download URL: [openai-2.31.0.tar.gz](https://files.pythonhosted.org/packages/94/fe/64b3d035780b3188f86c4f6f1bc202e7bb74757ef028802112273b9dcacf/openai-2.31.0.tar.gz) - Upload date: Apr 8, 2026 - Size: 684.8 kB - Tags: Source - Uploaded using Trusted Publishing? No - Uploaded via: twine/5.1.1 CPython/3.12.9 ### File hashes | Algorithm | Hash digest | | |---|---|---| | SHA256 | `43ca59a88fc973ad1848d86b98d7fac207e265ebbd1828b5e4bdfc85f79427a5` | Copy | | MD5 | `2687404315d7408b5753ae938fdbe688` | Copy | | BLAKE2b-256 | `94fe64b3d035780b3188f86c4f6f1bc202e7bb74757ef028802112273b9dcacf` | Copy | [See more details on using hashes here.](https://pip.pypa.io/en/stable/topics/secure-installs/#hash-checking-mode "External link") ## File details Details for the file `openai-2.31.0-py3-none-any.whl`. ### File metadata - Download URL: [openai-2.31.0-py3-none-any.whl](https://files.pythonhosted.org/packages/66/bc/a8f7c3aa03452fedbb9af8be83e959adba96a6b4a35e416faffcc959c568/openai-2.31.0-py3-none-any.whl) - Upload date: Apr 8, 2026 - Size: 1.2 MB - Tags: Python 3 - Uploaded using Trusted Publishing? No - Uploaded via: twine/5.1.1 CPython/3.12.9 ### File hashes | Algorithm | Hash digest | | |---|---|---| | SHA256 | `44e1344d87e56a493d649b17e2fac519d1368cbb0745f59f1957c4c26de50a0a` | Copy | | MD5 | `ffe6c7936c6c0cd292214c65d192fd79` | Copy | | BLAKE2b-256 | `66bca8f7c3aa03452fedbb9af8be83e959adba96a6b4a35e416faffcc959c568` | Copy | [See more details on using hashes here.](https://pip.pypa.io/en/stable/topics/secure-installs/#hash-checking-mode "External link") ![](https://pypi.org/static/images/white-cube.2351a86c.svg) ## Help - [Installing packages](https://packaging.python.org/tutorials/installing-packages/ "External link") - [Uploading packages](https://packaging.python.org/tutorials/packaging-projects/ "External link") - [User guide](https://packaging.python.org/ "External link") - [Project name retention](https://www.python.org/dev/peps/pep-0541/ "External link") - [FAQs](https://pypi.org/help/) ## About PyPI - [PyPI Blog](https://blog.pypi.org/ "External link") - [Infrastructure dashboard](https://dtdg.co/pypi "External link") - [Statistics](https://pypi.org/stats/) - [Logos & trademarks](https://pypi.org/trademarks/) - [Our sponsors](https://pypi.org/sponsors/) ## Contributing to PyPI - [Bugs and feedback](https://pypi.org/help/#feedback) - [Contribute on GitHub](https://github.com/pypi/warehouse "External link") - [Translate PyPI](https://hosted.weblate.org/projects/pypa/warehouse/ "External link") - [Sponsor PyPI](https://pypi.org/sponsors/) - [Development credits](https://github.com/pypi/warehouse/graphs/contributors "External link") ## Using PyPI - [Terms of Service](https://policies.python.org/pypi.org/Terms-of-Service/ "External link") - [Report security issue](https://pypi.org/security/) - [Code of conduct](https://policies.python.org/python.org/code-of-conduct/ "External link") - [Privacy Notice](https://policies.python.org/pypi.org/Privacy-Notice/ "External link") - [Acceptable Use Policy](https://policies.python.org/pypi.org/Acceptable-Use-Policy/ "External link") *** Status: [all systems operational](https://status.python.org/ "External link") Developed and maintained by the Python community, for the Python community. [Donate today\!](https://donate.pypi.org/) "PyPI", "Python Package Index", and the blocks logos are registered [trademarks](https://pypi.org/trademarks/) of the [Python Software Foundation](https://www.python.org/psf-landing). © 2026 [Python Software Foundation](https://www.python.org/psf-landing/ "External link") [Site map](https://pypi.org/sitemap/) Switch to desktop version Supported by [![](https://pypi-camo.freetls.fastly.net/ed7074cadad1a06f56bc520ad9bd3e00d0704c5b/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f6177732d77686974652d6c6f676f2d7443615473387a432e706e67) AWS Cloud computing and Security Sponsor](https://aws.amazon.com/) [![](https://pypi-camo.freetls.fastly.net/8855f7c063a3bdb5b0ce8d91bfc50cf851cc5c51/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f64617461646f672d77686974652d6c6f676f2d6668644c4e666c6f2e706e67) Datadog Monitoring](https://www.datadoghq.com/) [![](https://pypi-camo.freetls.fastly.net/60f709d24f3e4d469f9adc77c65e2f5291a3d165/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f6465706f742d77686974652d6c6f676f2d7038506f476831302e706e67) Depot Continuous Integration](https://depot.dev/) [![](https://pypi-camo.freetls.fastly.net/df6fe8829cbff2d7f668d98571df1fd011f36192/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f666173746c792d77686974652d6c6f676f2d65684d3077735f6f2e706e67) Fastly CDN](https://www.fastly.com/) [![](https://pypi-camo.freetls.fastly.net/420cc8cf360bac879e24c923b2f50ba7d1314fb0/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f676f6f676c652d77686974652d6c6f676f2d616734424e3774332e706e67) Google Download Analytics](https://careers.google.com/) [![](https://pypi-camo.freetls.fastly.net/d01053c02f3a626b73ffcb06b96367fdbbf9e230/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f70696e67646f6d2d77686974652d6c6f676f2d67355831547546362e706e67) Pingdom Monitoring](https://www.pingdom.com/) [![](https://pypi-camo.freetls.fastly.net/67af7117035e2345bacb5a82e9aa8b5b3e70701d/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f73656e7472792d77686974652d6c6f676f2d4a2d6b64742d706e2e706e67) Sentry Error logging](https://sentry.io/for/python/?utm_source=pypi&utm_medium=paid-community&utm_campaign=python-na-evergreen&utm_content=static-ad-pypi-sponsor-learnmore) [![](https://pypi-camo.freetls.fastly.net/b611884ff90435a0575dbab7d9b0d3e60f136466/68747470733a2f2f73746f726167652e676f6f676c65617069732e636f6d2f707970692d6173736574732f73706f6e736f726c6f676f732f737461747573706167652d77686974652d6c6f676f2d5467476c6a4a2d502e706e67) StatusPage Status page](https://statuspage.io/)
Readable Markdown
## OpenAI Python API library [![PyPI version](https://pypi-camo.freetls.fastly.net/594569836ed5abd55c2114c0e6795992b694fd4c/68747470733a2f2f696d672e736869656c64732e696f2f707970692f762f6f70656e61692e7376673f6c6162656c3d7079706925323028737461626c6529)](https://pypi.org/project/openai/) The OpenAI Python library provides convenient access to the OpenAI REST API from any Python 3.9+ application. The library includes type definitions for all request params and response fields, and offers both synchronous and asynchronous clients powered by [httpx](https://github.com/encode/httpx). It is generated from our [OpenAPI specification](https://github.com/openai/openai-openapi) with [Stainless](https://stainlessapi.com/). ## Documentation The REST API documentation can be found on [platform.openai.com](https://platform.openai.com/docs/api-reference). The full API of this library can be found in [api.md](https://github.com/openai/openai-python/tree/main/api.md). ## Installation ``` # install from PyPI pip install openai ``` ## Usage The full API of this library can be found in [api.md](https://github.com/openai/openai-python/tree/main/api.md). The primary API for interacting with OpenAI models is the [Responses API](https://platform.openai.com/docs/api-reference/responses). You can generate text from the model with the code below. ``` import os from openai import OpenAI client = OpenAI( # This is the default and can be omitted api_key=os.environ.get("OPENAI_API_KEY"), ) response = client.responses.create( model="gpt-5.2", instructions="You are a coding assistant that talks like a pirate.", input="How do I check if a Python object is an instance of a class?", ) print(response.output_text) ``` The previous standard (supported indefinitely) for generating text is the [Chat Completions API](https://platform.openai.com/docs/api-reference/chat). You can use that API to generate text from the model with the code below. ``` from openai import OpenAI client = OpenAI() completion = client.chat.completions.create( model="gpt-5.2", messages=[ {"role": "developer", "content": "Talk like a pirate."}, { "role": "user", "content": "How do I check if a Python object is an instance of a class?", }, ], ) print(completion.choices[0].message.content) ``` While you can provide an `api_key` keyword argument, we recommend using [python-dotenv](https://pypi.org/project/python-dotenv/) to add `OPENAI_API_KEY="My API Key"` to your `.env` file so that your API key is not stored in source control. [Get an API key here](https://platform.openai.com/settings/organization/api-keys). ### Workload Identity Authentication For secure, automated environments like cloud-managed Kubernetes, Azure, and Google Cloud Platform, you can use workload identity authentication with short-lived tokens from cloud identity providers instead of long-lived API keys. #### Kubernetes (service account tokens) ``` from openai import OpenAI from openai.auth import k8s_service_account_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": k8s_service_account_token_provider( "/var/run/secrets/kubernetes.io/serviceaccount/token" ), }, organization="org-xyz", project="proj-abc", ) response = client.chat.completions.create( model="gpt-4", messages=[{"role": "user", "content": "Hello!"}], ) ``` #### Azure (managed identity) ``` from openai import OpenAI from openai.auth import azure_managed_identity_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": azure_managed_identity_token_provider( resource="https://management.azure.com/", ), }, ) ``` #### Google Cloud Platform (compute engine metadata) ``` from openai import OpenAI from openai.auth import gcp_id_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": gcp_id_token_provider(audience="https://api.openai.com/v1"), }, ) ``` #### Custom subject token provider ``` from openai import OpenAI def get_custom_token() -> str: return "your-jwt-token" client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": { "token_type": "jwt", "get_token": get_custom_token, }, } ) ``` You can also customize the token refresh buffer (default is 1200 seconds (20 minutes) before expiration): ``` from openai import OpenAI from openai.auth import k8s_service_account_token_provider client = OpenAI( workload_identity={ "client_id": "your-client-id", "identity_provider_id": "idp-123", "service_account_id": "sa-456", "provider": k8s_service_account_token_provider("/var/token"), "refresh_buffer_seconds": 120.0, } ) ``` ### Vision With an image URL: ``` prompt = "What is in this image?" img_url = "https://upload.wikimedia.org/wikipedia/commons/thumb/d/d5/2023_06_08_Raccoon1.jpg/1599px-2023_06_08_Raccoon1.jpg" response = client.responses.create( model="gpt-5.2", input=[ { "role": "user", "content": [ {"type": "input_text", "text": prompt}, {"type": "input_image", "image_url": f"{img_url}"}, ], } ], ) ``` With the image as a base64 encoded string: ``` import base64 from openai import OpenAI client = OpenAI() prompt = "What is in this image?" with open("path/to/image.png", "rb") as image_file: b64_image = base64.b64encode(image_file.read()).decode("utf-8") response = client.responses.create( model="gpt-5.2", input=[ { "role": "user", "content": [ {"type": "input_text", "text": prompt}, {"type": "input_image", "image_url": f"data:image/png;base64,{b64_image}"}, ], } ], ) ``` ## Async usage Simply import `AsyncOpenAI` instead of `OpenAI` and use `await` with each API call: ``` import os import asyncio from openai import AsyncOpenAI client = AsyncOpenAI( # This is the default and can be omitted api_key=os.environ.get("OPENAI_API_KEY"), ) async def main() -> None: response = await client.responses.create( model="gpt-5.2", input="Explain disestablishmentarianism to a smart five year old." ) print(response.output_text) asyncio.run(main()) ``` Functionality between the synchronous and asynchronous clients is otherwise identical. ### With aiohttp By default, the async client uses `httpx` for HTTP requests. However, for improved concurrency performance you may also use `aiohttp` as the HTTP backend. You can enable this by installing `aiohttp`: ``` # install from PyPI pip install openai[aiohttp] ``` Then you can enable it by instantiating the client with `http_client=DefaultAioHttpClient()`: ``` import os import asyncio from openai import DefaultAioHttpClient from openai import AsyncOpenAI async def main() -> None: async with AsyncOpenAI( api_key=os.environ.get("OPENAI_API_KEY"), # This is the default and can be omitted http_client=DefaultAioHttpClient(), ) as client: chat_completion = await client.chat.completions.create( messages=[ { "role": "user", "content": "Say this is a test", } ], model="gpt-5.2", ) asyncio.run(main()) ``` ## Streaming responses We provide support for streaming responses using Server Side Events (SSE). ``` from openai import OpenAI client = OpenAI() stream = client.responses.create( model="gpt-5.2", input="Write a one-sentence bedtime story about a unicorn.", stream=True, ) for event in stream: print(event) ``` The async client uses the exact same interface. ``` import asyncio from openai import AsyncOpenAI client = AsyncOpenAI() async def main(): stream = await client.responses.create( model="gpt-5.2", input="Write a one-sentence bedtime story about a unicorn.", stream=True, ) async for event in stream: print(event) asyncio.run(main()) ``` ## Realtime API The Realtime API enables you to build low-latency, multi-modal conversational experiences. It currently supports text and audio as both input and output, as well as [function calling](https://platform.openai.com/docs/guides/function-calling) through a WebSocket connection. Under the hood the SDK uses the [`websockets`](https://websockets.readthedocs.io/en/stable/) library to manage connections. The Realtime API works through a combination of client-sent events and server-sent events. Clients can send events to do things like update session configuration or send text and audio inputs. Server events confirm when audio responses have completed, or when a text response from the model has been received. A full event reference can be found [here](https://platform.openai.com/docs/api-reference/realtime-client-events) and a guide can be found [here](https://platform.openai.com/docs/guides/realtime). Basic text based example: ``` import asyncio from openai import AsyncOpenAI async def main(): client = AsyncOpenAI() async with client.realtime.connect(model="gpt-realtime") as connection: await connection.session.update( session={"type": "realtime", "output_modalities": ["text"]} ) await connection.conversation.item.create( item={ "type": "message", "role": "user", "content": [{"type": "input_text", "text": "Say hello!"}], } ) await connection.response.create() async for event in connection: if event.type == "response.output_text.delta": print(event.delta, flush=True, end="") elif event.type == "response.output_text.done": print() elif event.type == "response.done": break asyncio.run(main()) ``` However the real magic of the Realtime API is handling audio inputs / outputs, see this example [TUI script](https://github.com/openai/openai-python/blob/main/examples/realtime/push_to_talk_app.py) for a fully fledged example. ### Realtime error handling Whenever an error occurs, the Realtime API will send an [`error` event](https://platform.openai.com/docs/guides/realtime-model-capabilities#error-handling) and the connection will stay open and remain usable. This means you need to handle it yourself, as *no errors are raised directly* by the SDK when an `error` event comes in. ``` client = AsyncOpenAI() async with client.realtime.connect(model="gpt-realtime") as connection: ... async for event in connection: if event.type == 'error': print(event.error.type) print(event.error.code) print(event.error.event_id) print(event.error.message) ``` ## Using types Nested request parameters are [TypedDicts](https://docs.python.org/3/library/typing.html#typing.TypedDict). Responses are [Pydantic models](https://docs.pydantic.dev/) which also provide helper methods for things like: - Serializing back into JSON, `model.to_json()` - Converting to a dictionary, `model.to_dict()` Typed requests and responses provide autocomplete and documentation within your editor. If you would like to see type errors in VS Code to help catch bugs earlier, set `python.analysis.typeCheckingMode` to `basic`. ## Pagination List methods in the OpenAI API are paginated. This library provides auto-paginating iterators with each list response, so you do not have to request successive pages manually: ``` from openai import OpenAI client = OpenAI() all_jobs = [] # Automatically fetches more pages as needed. for job in client.fine_tuning.jobs.list( limit=20, ): # Do something with job here all_jobs.append(job) print(all_jobs) ``` Or, asynchronously: ``` import asyncio from openai import AsyncOpenAI client = AsyncOpenAI() async def main() -> None: all_jobs = [] # Iterate through items across all pages, issuing requests as needed. async for job in client.fine_tuning.jobs.list( limit=20, ): all_jobs.append(job) print(all_jobs) asyncio.run(main()) ``` Alternatively, you can use the `.has_next_page()`, `.next_page_info()`, or `.get_next_page()` methods for more granular control working with pages: ``` first_page = await client.fine_tuning.jobs.list( limit=20, ) if first_page.has_next_page(): print(f"will fetch next page using these details: {first_page.next_page_info()}") next_page = await first_page.get_next_page() print(f"number of items we just fetched: {len(next_page.data)}") # Remove `await` for non-async usage. ``` Or just work directly with the returned data: ``` first_page = await client.fine_tuning.jobs.list( limit=20, ) print(f"next page cursor: {first_page.after}") # => "next page cursor: ..." for job in first_page.data: print(job.id) # Remove `await` for non-async usage. ``` ## Nested params Nested parameters are dictionaries, typed using `TypedDict`, for example: ``` from openai import OpenAI client = OpenAI() response = client.chat.responses.create( input=[ { "role": "user", "content": "How much ?", } ], model="gpt-5.2", response_format={"type": "json_object"}, ) ``` ## File uploads Request parameters that correspond to file uploads can be passed as `bytes`, or a [`PathLike`](https://docs.python.org/3/library/os.html#os.PathLike) instance or a tuple of `(filename, contents, media type)`. ``` from pathlib import Path from openai import OpenAI client = OpenAI() client.files.create( file=Path("input.jsonl"), purpose="fine-tune", ) ``` The async client uses the exact same interface. If you pass a [`PathLike`](https://docs.python.org/3/library/os.html#os.PathLike) instance, the file contents will be read asynchronously automatically. ## Webhook Verification Verifying webhook signatures is *optional but encouraged*. For more information about webhooks, see [the API docs](https://platform.openai.com/docs/guides/webhooks). ### Parsing webhook payloads For most use cases, you will likely want to verify the webhook and parse the payload at the same time. To achieve this, we provide the method `client.webhooks.unwrap()`, which parses a webhook request and verifies that it was sent by OpenAI. This method will raise an error if the signature is invalid. Note that the `body` parameter must be the raw JSON string sent from the server (do not parse it first). The `.unwrap()` method will parse this JSON for you into an event object after verifying the webhook was sent from OpenAI. ``` from openai import OpenAI from flask import Flask, request app = Flask(__name__) client = OpenAI() # OPENAI_WEBHOOK_SECRET environment variable is used by default @app.route("/webhook", methods=["POST"]) def webhook(): request_body = request.get_data(as_text=True) try: event = client.webhooks.unwrap(request_body, request.headers) if event.type == "response.completed": print("Response completed:", event.data) elif event.type == "response.failed": print("Response failed:", event.data) else: print("Unhandled event type:", event.type) return "ok" except Exception as e: print("Invalid signature:", e) return "Invalid signature", 400 if __name__ == "__main__": app.run(port=8000) ``` ### Verifying webhook payloads directly In some cases, you may want to verify the webhook separately from parsing the payload. If you prefer to handle these steps separately, we provide the method `client.webhooks.verify_signature()` to *only verify* the signature of a webhook request. Like `.unwrap()`, this method will raise an error if the signature is invalid. Note that the `body` parameter must be the raw JSON string sent from the server (do not parse it first). You will then need to parse the body after verifying the signature. ``` import json from openai import OpenAI from flask import Flask, request app = Flask(__name__) client = OpenAI() # OPENAI_WEBHOOK_SECRET environment variable is used by default @app.route("/webhook", methods=["POST"]) def webhook(): request_body = request.get_data(as_text=True) try: client.webhooks.verify_signature(request_body, request.headers) # Parse the body after verification event = json.loads(request_body) print("Verified event:", event) return "ok" except Exception as e: print("Invalid signature:", e) return "Invalid signature", 400 if __name__ == "__main__": app.run(port=8000) ``` ## Handling errors When the library is unable to connect to the API (for example, due to network connection problems or a timeout), a subclass of `openai.APIConnectionError` is raised. When the API returns a non-success status code (that is, 4xx or 5xx response), a subclass of `openai.APIStatusError` is raised, containing `status_code` and `response` properties. All errors inherit from `openai.APIError`. ``` import openai from openai import OpenAI client = OpenAI() try: client.fine_tuning.jobs.create( model="gpt-4o", training_file="file-abc123", ) except openai.APIConnectionError as e: print("The server could not be reached") print(e.__cause__) # an underlying Exception, likely raised within httpx. except openai.RateLimitError as e: print("A 429 status code was received; we should back off a bit.") except openai.APIStatusError as e: print("Another non-200-range status code was received") print(e.status_code) print(e.response) ``` Error codes are as follows: | Status Code | Error Type | |---|---| | 400 | `BadRequestError` | | 401 | `AuthenticationError` | | 403 | `PermissionDeniedError` | | 404 | `NotFoundError` | | 422 | `UnprocessableEntityError` | | 429 | `RateLimitError` | | \>=500 | `InternalServerError` | | N/A | `APIConnectionError` | ## Request IDs > For more information on debugging requests, see [these docs](https://platform.openai.com/docs/api-reference/debugging-requests) All object responses in the SDK provide a `_request_id` property which is added from the `x-request-id` response header so that you can quickly log failing requests and report them back to OpenAI. ``` response = await client.responses.create( model="gpt-5.2", input="Say 'this is a test'.", ) print(response._request_id) # req_123 ``` Note that unlike other properties that use an `_` prefix, the `_request_id` property *is* public. Unless documented otherwise, *all* other `_` prefix properties, methods and modules are *private*. > \[!IMPORTANT\] > If you need to access request IDs for failed requests you must catch the `APIStatusError` exception ``` import openai try: completion = await client.chat.completions.create( messages=[{"role": "user", "content": "Say this is a test"}], model="gpt-5.2" ) except openai.APIStatusError as exc: print(exc.request_id) # req_123 raise exc ``` ## Retries Certain errors are automatically retried 2 times by default, with a short exponential backoff. Connection errors (for example, due to a network connectivity problem), 408 Request Timeout, 409 Conflict, 429 Rate Limit, and \>=500 Internal errors are all retried by default. You can use the `max_retries` option to configure or disable retry settings: ``` from openai import OpenAI # Configure the default for all requests: client = OpenAI( # default is 2 max_retries=0, ) # Or, configure per-request: client.with_options(max_retries=5).chat.completions.create( messages=[ { "role": "user", "content": "How can I get the name of the current day in JavaScript?", } ], model="gpt-5.2", ) ``` ## Timeouts By default requests time out after 10 minutes. You can configure this with a `timeout` option, which accepts a float or an [`httpx.Timeout`](https://www.python-httpx.org/advanced/timeouts/#fine-tuning-the-configuration) object: ``` from openai import OpenAI # Configure the default for all requests: client = OpenAI( # 20 seconds (default is 10 minutes) timeout=20.0, ) # More granular control: client = OpenAI( timeout=httpx.Timeout(60.0, read=5.0, write=10.0, connect=2.0), ) # Override per-request: client.with_options(timeout=5.0).chat.completions.create( messages=[ { "role": "user", "content": "How can I list all files in a directory using Python?", } ], model="gpt-5.2", ) ``` On timeout, an `APITimeoutError` is thrown. Note that requests that time out are [retried twice by default](https://github.com/openai/openai-python/tree/main/#retries). ## Advanced ### Logging We use the standard library [`logging`](https://docs.python.org/3/library/logging.html) module. You can enable logging by setting the environment variable `OPENAI_LOG` to `info`. ``` $ export OPENAI_LOG=info ``` Or to `debug` for more verbose logging. ### How to tell whether `None` means `null` or missing In an API response, a field may be explicitly `null`, or missing entirely; in either case, its value is `None` in this library. You can differentiate the two cases with `.model_fields_set`: ``` if response.my_field is None: if 'my_field' not in response.model_fields_set: print('Got json like {}, without a "my_field" key present at all.') else: print('Got json like {"my_field": null}.') ``` ### Accessing raw response data (e.g. headers) The "raw" Response object can be accessed by prefixing `.with_raw_response.` to any HTTP method call, e.g., ``` from openai import OpenAI client = OpenAI() response = client.chat.completions.with_raw_response.create( messages=[{ "role": "user", "content": "Say this is a test", }], model="gpt-5.2", ) print(response.headers.get('X-My-Header')) completion = response.parse() # get the object that `chat.completions.create()` would have returned print(completion) ``` These methods return a [`LegacyAPIResponse`](https://github.com/openai/openai-python/tree/main/src/openai/_legacy_response.py) object. This is a legacy class as we're changing it slightly in the next major version. For the sync client this will mostly be the same with the exception of `content` & `text` will be methods instead of properties. In the async client, all methods will be async. A migration script will be provided & the migration in general should be smooth. #### `.with_streaming_response` The above interface eagerly reads the full response body when you make the request, which may not always be what you want. To stream the response body, use `.with_streaming_response` instead, which requires a context manager and only reads the response body once you call `.read()`, `.text()`, `.json()`, `.iter_bytes()`, `.iter_text()`, `.iter_lines()` or `.parse()`. In the async client, these are async methods. As such, `.with_streaming_response` methods return a different [`APIResponse`](https://github.com/openai/openai-python/tree/main/src/openai/_response.py) object, and the async client returns an [`AsyncAPIResponse`](https://github.com/openai/openai-python/tree/main/src/openai/_response.py) object. ``` with client.chat.completions.with_streaming_response.create( messages=[ { "role": "user", "content": "Say this is a test", } ], model="gpt-5.2", ) as response: print(response.headers.get("X-My-Header")) for line in response.iter_lines(): print(line) ``` The context manager is required so that the response will reliably be closed. ### Making custom/undocumented requests This library is typed for convenient access to the documented API. If you need to access undocumented endpoints, params, or response properties, the library can still be used. #### Undocumented endpoints To make requests to undocumented endpoints, you can make requests using `client.get`, `client.post`, and other http verbs. Options on the client will be respected (such as retries) when making this request. ``` import httpx response = client.post( "/foo", cast_to=httpx.Response, body={"my_param": True}, ) print(response.headers.get("x-foo")) ``` #### Undocumented request params If you want to explicitly send an extra param, you can do so with the `extra_query`, `extra_body`, and `extra_headers` request options. #### Undocumented response properties To access undocumented response properties, you can access the extra fields like `response.unknown_prop`. You can also get all the extra fields on the Pydantic model as a dict with [`response.model_extra`](https://docs.pydantic.dev/latest/api/base_model/#pydantic.BaseModel.model_extra). ### Configuring the HTTP client You can directly override the [httpx client](https://www.python-httpx.org/api/#client) to customize it for your use case, including: - Support for [proxies](https://www.python-httpx.org/advanced/proxies/) - Custom [transports](https://www.python-httpx.org/advanced/transports/) - Additional [advanced](https://www.python-httpx.org/advanced/clients/) functionality ``` import httpx from openai import OpenAI, DefaultHttpxClient client = OpenAI( # Or use the `OPENAI_BASE_URL` env var base_url="http://my.test.server.example.com:8083/v1", http_client=DefaultHttpxClient( proxy="http://my.test.proxy.example.com", transport=httpx.HTTPTransport(local_address="0.0.0.0"), ), ) ``` You can also customize the client on a per-request basis by using `with_options()`: ``` client.with_options(http_client=DefaultHttpxClient(...)) ``` ### Managing HTTP resources By default the library closes underlying HTTP connections whenever the client is [garbage collected](https://docs.python.org/3/reference/datamodel.html#object.__del__). You can manually close the client using the `.close()` method if desired, or with a context manager that closes when exiting. ``` from openai import OpenAI with OpenAI() as client: # make requests here ... # HTTP client is now closed ``` ## Microsoft Azure OpenAI To use this library with [Azure OpenAI](https://learn.microsoft.com/azure/ai-services/openai/overview), use the `AzureOpenAI` class instead of the `OpenAI` class. > \[!IMPORTANT\] The Azure API shape differs from the core API shape which means that the static types for responses / params won't always be correct. ``` from openai import AzureOpenAI # gets the API Key from environment variable AZURE_OPENAI_API_KEY client = AzureOpenAI( # https://learn.microsoft.com/azure/ai-services/openai/reference#rest-api-versioning api_version="2023-07-01-preview", # https://learn.microsoft.com/azure/cognitive-services/openai/how-to/create-resource?pivots=web-portal#create-a-resource azure_endpoint="https://example-endpoint.openai.azure.com", ) completion = client.chat.completions.create( model="deployment-name", # e.g. gpt-35-instant messages=[ { "role": "user", "content": "How do I output all files in a directory using Python?", }, ], ) print(completion.to_json()) ``` In addition to the options provided in the base `OpenAI` client, the following options are provided: - `azure_endpoint` (or the `AZURE_OPENAI_ENDPOINT` environment variable) - `azure_deployment` - `api_version` (or the `OPENAI_API_VERSION` environment variable) - `azure_ad_token` (or the `AZURE_OPENAI_AD_TOKEN` environment variable) - `azure_ad_token_provider` An example of using the client with Microsoft Entra ID (formerly known as Azure Active Directory) can be found [here](https://github.com/openai/openai-python/blob/main/examples/azure_ad.py). ## Versioning This package generally follows [SemVer](https://semver.org/spec/v2.0.0.html) conventions, though certain backwards-incompatible changes may be released as minor versions: 1. Changes that only affect static types, without breaking runtime behavior. 2. Changes to library internals which are technically public but not intended or documented for external use. *(Please open a GitHub issue to let us know if you are relying on such internals.)* 3. Changes that we do not expect to impact the vast majority of users in practice. We take backwards-compatibility seriously and work hard to ensure you can rely on a smooth upgrade experience. We are keen for your feedback; please open an [issue](https://www.github.com/openai/openai-python/issues) with questions, bugs, or suggestions. ### Determining the installed version If you've upgraded to the latest version but aren't seeing any new features you were expecting then your python environment is likely still using an older version. You can determine the version that is being used at runtime with: ``` import openai print(openai.__version__) ``` ## Requirements Python 3.9 or higher. ## Contributing See [the contributing documentation](https://github.com/openai/openai-python/tree/main/CONTRIBUTING.md).
Shard59 (laksa)
Root Hash7813724874982801459
Unparsed URLorg,pypi!/project/openai/ s443