NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI, #97 most downloaded on PyPI
The official Python library for the openai API
Latest stable
3.57.0
Last release 2 days ago
20 Nov 2026
Ships on a steady schedule
a new release about every 9 days
Nearly every release is documented
notes for 60 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
7 years old
478 releases · first in 2020
Drag across the ruler to compare, or tap a release to open it.
api: support Live WebRTC standby
api: add gpt-live-1.1 to Live model choices
Your coding agent can read these notes before it upgrades. Set up the MCP server →
api: document CSAM moderation response fields
api: type Managed Agents budget events
api: add 2-hour and 8-hour Responses background windows
api: expand agent, image, Live, and external key support
api: expose reasoning input transformations on responses
api: add stream options for agent sessions
api: add Australia, India, and Japan residency options
api: add 12-hour Responses caching and cache-write usage
api: add workload identity providers and federation rules
api: expand Live Responses delegation tools
api: preserve Responses WebSocket connection behavior
api: add agent context compaction settings
api: accept typed references in Realtime response inputs
api: add prompt cache breakpoint TTL options
api: add gpt-6.1-terra model identifier
api: add delegation controls and prompt cache lifetimes
api: This minor release includes a breaking change to Decisions types. In Live response models, question instructions can now be a string or a list of…
instructions can now be a string or a list of structured text, image, or JSON parts; check isinstance(question.instructions, str) before using string operations and handle the structured form. Decisions request input types also include additional message roles and tool-history items, so update exhaustive type checks accordingly. Existing string instructions and DecisionInputMessageParam user-message inputs remain supported.api: honor per-call WebSocket query settings
add Vertex AI support for the Responses API
preserve beta Realtime byte reads on websockets 13
api: add typed Live sideband events
api: add cache settings and key options [2/2]
add browser and computer toolsets (1/4)
api: add beta session spending limits
api: add billing_not_active session error code
api: add agent environment suspension and expiration
api: add prewarmed hosted environments
api: correct custom voice creation parameters
api: add standalone Decisions support
agents: observe local tool failures
api: add custom voice creation and agent session events
agents: [1/n] return typed answers from session streams
api: correct the missing authentication error message ( #3993 ) ( 51692ed ), closes #3962
api: add computer use to beta agents
api: add GPT-6.1 Sol model identifier
api: add Agents credential and session options
preserve single files for fallback extraction paths
chat: preserve single-pass tool iterables
api: add GCP external storage support
api: add GPT-6 Sol and Luna model identifiers
lib: use log.warning instead of the deprecated log.warn
parsing: drop TextFormatT parameterization in parse_response to fix memory leak
api: avoid loading unrelated API resources on first use
api: add agent session model settings
client: validate retry limits and preserve application errors
streaming: normalize errors raised while reading streams
3.13.0 (2026-09-10) Features api: add Agents API
add aclose() to AsyncStream for standard async cleanup
api: Add expiration controls for service account keys
api: Add prompt cache diagnostics
One column per quarter.