NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #98 most downloaded on PyPI
The official Python library for the openai API
Last release today
02 Oct 2026
Ships on a steady schedule
a new release about every 2 weeks
Nearly every release is documented
notes for 60 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
7 years old
441 releases · first in 2020
One column per quarter.
readme: fix misleading timeout example value
## 1.25.1 (2024-05-02) Full Changelog: v1.25.0...v1.25.1 ### Chores * internal: bump prism version
## 1.25.0 (2024-05-01) Full Changelog: v1.24.1...v1.25.0 ### Features * api: delete messages
internal: add link to openapi spec
client: log response headers in debug mode
internal: update test helper function
internal: use actions/checkout@v4 for codeflow
api: change timestamps to unix integers
## 1.23.3 (2024-04-23) Full Changelog: v1.23.2...v1.23.3 ### Chores * internal: restructure imports
api: correct types for message attachment tools
api: correct types for attachments
api: add request id property to response classes
internal: ban usage of lru_cache
internal: add lru_cache helper function
api: docs and response_format response property
## 1.21.0 (2024-04-17) Full Changelog: v1.20.0...v1.21.0 ### Features * api: add vector stores
client: add header OpenAI-Project
errors: add request_id property
## 1.18.0 (2024-04-15) Full Changelog: v1.17.1...v1.18.0 ### Features * api: add batch API (#1316) (3e6f19e) * api: updates
## 1.17.1 (2024-04-12) Full Changelog: v1.17.0...v1.17.1 ### Chores * fix typo (#1304) (1129082) * internal: formatting
api: add additional messages when creating thread run
client: correct logic for line decoding in streaming
internal: defer model build for import latency
api: add support for filtering messages by run_id
revert regression with 3.7 support
internal: update generated pragma comment
Full Changelog: v1.14.0...v1.14.1 ### Documentation * readme: assistant streaming
assistants: add support for streaming
streaming: improve error messages
Full Changelog: v1.13.3...v1.13.4
api: add wav and pcm to response_format
api: add timestamp_granularities, add gpt-3.5-turbo-0125 model
Full Changelog: v1.11.1...v1.12.0
timestamp_granularities, add gpt-3.5-turbo-0125 model (#1125) (1ecf8f6)--model arg (#1132) (0d53866)prevent crash when platform.architecture() is not allowed
Full Changelog: v1.11.0...v1.11.1
client: support parsing custom response types
Full Changelog: v1.10.0...v1.11.0
api: add text embeddings dimensions param
api: add usage to runs and run steps
client: add support for streaming raw responses
Full Changelog: v1.7.2...v1.8.0
api: remove deprecated endpoints
client: improve debug logging for failed requests
Full Changelog: v1.7.0...v1.7.1
add None default value to nullable response properties
Full Changelog: v1.6.1...v1.7.0
internal: use ruff instead of black for formatting
Full Changelog: v1.6.0...v1.6.1
api: add additional instructions for runs
Full Changelog: v1.5.0...v1.6.0
api: add token logprobs to chat completions
api: add optional name argument + improve docs
Full Changelog: v1.3.9...v1.4.0
small Improvement in the async chat response code
avoid leaking memory when Client.with_options is used
Full Changelog: v1.3.7...v1.3.8
client: correct base_url setter implementation
Full Changelog: v1.3.6...v1.3.7
client: add support for streaming binary responses
azure: ensure custom options can be passed to copy
client: attempt to parse unknown json content types
internal: update type hint for helper function
Full Changelog: v1.3.2...v1.3.3
Full Changelog: v1.3.1...v1.3.2 ### Documentation * readme: minor updates
Full Changelog: v1.3.0...v1.3.1 ### Chores * internal: add publish script
client: support reading the base url from an env variable
Full Changelog: v1.2.4...v1.3.0
client: retry if SSLWantReadError occurs in the async client
Full Changelog: v1.2.3...v1.2.4
cli/audio: file format detection failing for whisper
Full Changelog: v1.2.2...v1.2.3
client: correctly assign error properties
client: deprecate files.retrieve_content in favour of files.content
client: support passing chunk size for binary responses
Full Changelog: v1.1.2...v1.2.0
api: accidentally required params, add new models & other fixes
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →