v0.6.1

What's Changed

• fix: invalid model setting when passing prompt to Agent by @seratch in #1852
• docs: document input guardrail execution modes by @RVV-karma in #2084
• Update all translated document pages by @github-actions[bot] in #2104
• fix: #2097 fix the default settings for SIP realtime runner by @seratch in #2105
• version 0.6.1 by @seratch in #2106

Full Changelog: v0.6.0...v0.6.1

v0.6.0

Key changes

Handoff behavior has a breaking change in this version. Specifically, the message history is now collapsed into a single message when handing off to a new agent. We have verified and evaluated this behavior and think it works better, but recommend testing with your agents before upgrading to v0.6.0 in production.

What's Changed

• fix(twilio): add configurable startup delay to avoid initial audio jitter (fixes #1906) by @Shraman123 in #2033
• fix: #2081 Redis SSL Connection Error with redis-py v7 by @yju0808 in #2082
• Nest handoff history by default by @jhills20 in #1996
• Upgrade openai-python to 2.8.0 for GPT 5.1 by @ihower in #2091
• Update all translated document pages by @github-actions[bot] in #2092
• Add prompt_cache_retention to ModelSettings by @ihower in #2095
• feat: add run_in_parallel parameter to input guardrails by @RVV-karma in #1986
• fix: #2008 Fix agent memory leak using weakref by @DanieleMorotti in #2014
• docs: fix typos in some files by @Edge-Seven in #2096
• feat: add tool error logging by @danilbogdan in #1804
• v0.6.0 by @rm-openai in #2093

New Contributors

• @Shraman123 made their first contribution in #2033
• @yju0808 made their first contribution in #2082
• @RVV-karma made their first contribution in #1986
• @Edge-Seven made their first contribution in #2096
• @danilbogdan made their first contribution in #1804

Full Changelog: v0.5.1...v0.6.0

v0.5.1

Key update is support for new tools from the GPT-5.1 launch: shell and apply_patch.

What's Changed

• Docs: Add per-request usage to docs and example script by @habema in #2026
• docs: Update document pages for v0.5.0 release by @seratch in #2019
• Update all translated document pages by @github-actions[bot] in #2044
• Fix Anthropic model used in LiteLLM example by @amri369 in #2047
• fix: #2073 Improve type hinting in memory extension modules by @habema in #2077
• Add new tools for gpt-5.1 by @dkundel-openai in #2079
• v0.5.1 by @rm-openai in #2080

Full Changelog: v0.5.0...v0.5.1

v0.5.0

Key Changes

This version doesn’t introduce any visible breaking changes, but it includes new features and a few significant updates under the hood:

• Added support for RealtimeRunner to handle SIP protocol connections
• Significantly revised the internal logic of Runner#run_sync for Python 3.14 compatibility

What's Changed

• docs: #1999 adjust the realtime module code comment by @seratch in #2000
• docs: improve typing in lifecycle hook example code by @seratch in #2002
• Add Python 3.14 to the supported versions by @seratch in #1961
• feat: #1760 Add SIP support for realtime agent runner by @seratch in #1993
• fix: improve run_sync's python 3.14 compatibility by @seratch in #2006
• feat: #1994 Add a list of per-request usage data to Usage by @habema in #1995
• fix: use ResponseOutputRefusal for refusal content part added event by @mshsheikh in #2013
• docs: clarify on_tool_start and on_tool_end hooks for local tools only by @ihower in #2037
• fix: #2038 openai==2.7.0 compatibility by @seratch in #2039
• fix(usage): Normalize None token details on Usage initialization by @codefromthecrypt in #2034
• fix(realtime): include tool arguments in RealtimeToolStart/RealtimeToolEnd events by @ruskaruma in #2028
• Add Dapr session storage option by @filintod in #1937
• v0.5.0 by @seratch in #2015

New Contributors

• @codefromthecrypt made their first contribution in #2034

Full Changelog: v0.4.2...v0.5.0

v0.4.2

What's Changed

• fix: #1764 enable developers to pass custom reasoning effort for LiteLLM providers by @seratch in #1972
• ci: make format required (devs should run make format before committing) by @filintod in #1979
• Fix: Add empty checks for reasoning content arrays in stream handler by @gn00295120 in #1982
• fix: #1942 Enable async tool calling in Realtime sessions by @seratch in #1984
• Fix #1559: Handle empty choices array in LiteLLM model (same as PR #935) by @gn00295120 @ihower in #1981
• v0.4.2 by @seratch in #1988

Full Changelog: v0.4.1...v0.4.2

v0.4.1

What's Changed

• Add breaking change notes in docs by @seratch in #1894
• Update all translated document pages by @github-actions[bot] in #1922
• docs: add missing items in side navi bar (ja, zh) by @seratch in #1932
• Update all translated document pages by @github-actions[bot] in #1933
• Fix: call _fix_tool_message_ordering for all Anthropic models by @habema in #1936
• make format on main branch returns diff on some files by @filintod in #1947
• fix: prevent ValueError when audio buffer is empty in STT (fixes #821) by @gn00295120 in #1934
• fix: add missing mcp_approval_response event name (fixes #1020) by @gn00295120 in #1931
• fix: Remove trailing commas causing tuple assignment in response cancellation by @gn00295120 in #1952
• fix: await cancelled websocket task to prevent resource leak by @gn00295120 in #1955
• fix: prevent race condition in listener iteration by @gn00295120 in #1956
• Tweak CI test job settings by @seratch in #1962
• Remove unused ui.py example file by @ihower in #1967
• fix: #1907 guardrails w/ turn_detection.interrupt_response: true by @seratch in #1968
• Fix #1846 Litellm: fails with function name for tool_choice parameter w/ streaming enabled by @ihower in #1971
• Fix streaming trace end before guardrails complete by @zbirenbaum in #1921
• Add engine accessor to SQLAlchemySession for closing it when it's created from a URL by @gn00295120 in #1960
• Make Tool Output dict conversion stricter to improve backward compatibility by @ihower in #1965
• version 0.4.1 by @seratch in #1963

New Contributors

• @filintod made their first contribution in #1947
• @gn00295120 made their first contribution in #1934
• @zbirenbaum made their first contribution in #1921

Full Changelog: v0.4.0...v0.4.1

v0.4.0

Key Changes

In this version, openai package v1.x versions are no longer supported. Please use openai v2.x along with this SDK.

What's Changed

• fix: making input_file's filename optional in code for supporting non-OpenAI models by @ant31 in #1839
• feat: #1829 add httpx_client_factory to MCPServerStreamableHttp initialization options by @jjcampana in #1833
• fix: #1789 Fix multi-turn handling for conversation_id and previous_response_id: only send new items by @ihower in #1827
• Update all translated document pages by @github-actions[bot] in #1842
• docs: improve formatting consistency in tools documentation by @AbdulSamad94 in #1856
• Update all translated document pages by @github-actions[bot] in #1858
• docs: add documentation for extension feature EncryptedSession by @maxmekiska in #1872
• Update all translated document pages by @github-actions[bot] in #1873
• Fix: Correct streaming order for ReasoningItem and RawResponsesStreamEvent events by @ihower in #1869
• fix: adjust realtime handoff import by @mcbortolozzo in #1861
• Migrate openai from 1.x to 2.2.0 by @seratch in #1874
• fix: #1840 roll back session changes when Guardrail tripwire is triggered by @seratch in #1843
• docs: Add Chinese translation for documents by @SepineTam in #1878
• docs: add ToolContext section for advanced tool metadata by @MuhammadSaqib786 in #1868
• Fix: Exclude unset fields in OpenAIConversationsSession.get_items() by @ihower in #1883
• Update all translated document pages by @github-actions[bot] in #1891
• Feat: Expose MCP message handler configuration by @freeznet in #1795
• fix: handle omit type in _remove_not_given() for litellm model by @HareeshBahuleyan in #1893
• fix: convert oneOf to anyOf in strict schema for OpenAI compatibility by @ruskaruma in #1884
• Fix Local shell tool: return tool output to the LLM by @ihower in #1855
• Image, file output types for functions by @rm-openai in #1898
• Update all translated document pages by @github-actions[bot] in #1910
• Add graceful cancel mode for streaming runs by @habema in #1896
• Document AdvancedSQLiteSession and refactor session docs by @habema in #1791
• Update all translated document pages by @github-actions[bot] in #1913
• Fixes realtime example app stability issues by @thli-openai in #1905
• fix: #1900 fix a bug where SQLAlchemySession could return items in an invalid order by @seratch in #1917
• fix: #1916 openai_chatcompletions.Converter.extract_all_content does not support input_audio type items by @seratch in #1918
• v0.4.0 by @seratch in #1919

New Contributors

• @jjcampana made their first contribution in #1833
• @mcbortolozzo made their first contribution in #1861
• @SepineTam made their first contribution in #1878
• @MuhammadSaqib786 made their first contribution in #1868
• @freeznet made their first contribution in #1795
• @HareeshBahuleyan made their first contribution in #1893
• @ruskaruma made their first contribution in #1884
• @thli-openai made their first contribution in #1905

Full Changelog: v0.3.3...v0.4.0

v0.3.3

What's Changed

• Feat: Add AdvancedSQLiteSession with conversation branching & usage tracking by @habema in #1662
• Fix extended thinking bug by reordering tool messages by @habema in #1798
• feat: add Redis session support for scalable distributed memory by @damianoneill in #1785
• Add agent hooks to lifecycle_example by @seratch in #1809
• Update models in docs by @seratch in #1810
• Update all translated document pages by @github-actions[bot] in #1811
• Fix code snippet error in MCP docs by @seratch in #1812
• Update all translated document pages by @github-actions[bot] in #1813
• Adding tool input and output guardrails by @steven10a in #1792
• Korean translation by @tylerryu-oai in #1816
• Update all translated document pages by @github-actions[bot] in #1818
• docs: update examples.md to match current examples directory structure by @AbdulSamad94 in #1819
• Update all translated document pages by @github-actions[bot] in #1820
• Add guidelines for translation script contributors by @seratch in #1821
• Add missing parens in docs links by @tcbegley in #1825
• Update all translated document pages by @github-actions[bot] in #1826
• docs: fix link errors in sessions document page by @seratch in #1835
• Update all translated document pages by @github-actions[bot] in #1836
• fix: #1834 minor error in docstring by @seratch in #1837
• v0.3.3 by @rm-openai in #1838

New Contributors

• @damianoneill made their first contribution in #1785
• @steven10a made their first contribution in #1792
• @tylerryu-oai made their first contribution in #1816
• @AbdulSamad94 made their first contribution in #1819
• @tcbegley made their first contribution in #1825

Full Changelog: v0.3.2...v0.3.3

v0.3.2

What's Changed

• fix #1750 better error message when passing AgentHooks to Runner by @seratch in #1752
• Enhance function tool schemas with Annotated types by @MuhammadHassaanArain in #1777
• fix(chatcmpl): preserve all LiteLLM thinking-block signatures in converter by @serialx in #1779
• Update dynamic_system_prompt.py example to use dataclass for simiplicity by @SheikhMuhammadHamza-HS in #1774
• fix(chatcmpl): preserve text content when adding Anthropic thinking blocks to tool calls by @serialx in #1784
• fix: #556 add llms.txt / llms-full.txt by @seratch in #1782
• Add tool call arguments in ToolContext for RunHooks by @ihower in #1765
• Allow headers override instead of just ua by @jiwon-oai in #1787
• v0.3.2 by @rm-openai in #1793

New Contributors

• @SheikhMuhammadHamza-HS made their first contribution in #1774

Full Changelog: v0.3.1...v0.3.2

v0.3.1

What's Changed

• Update all translated document pages by @github-actions[bot] in #1720
• Fix missing imports in MCP documentation examples by @MariumAfzal2703 in #1718
• Update all translated document pages by @github-actions[bot] in #1721
• fix(streaming): push processed_response.new_items (including HandoffCallItem) to event_queue by @Wujiaxuan007 in #1703
• fix: #1704 Preserve thinking blocks in Anthropic conversations with tool calls by @habema in #1706
• Wait for input guardrails in streaming runs by @rm-openai in #1730
• docs: add docstring for parameter session in run, run_sync and run_streamed functions by @LokinDev in #1739
• docs: #1722 LiteLLM usage tracking by @seratch in #1743
• Update all translated document pages by @github-actions[bot] in #1747
• docs: Fix tracing setup documentation by @abdul-kabir-jawed in #1734
• Fixed Race Condition in RunResultStreaming.stream_events() Method by @AnkanMisra in #1745
• feat: Support Anthropic extended thinking and interleaved thinking by @serialx in #1744
• feat(realtime): #1560 add input audio noise reduction by @notV3NOM in #1749
• feat(voice)!: migrate STT streaming to match GA Realtime API by @notV3NOM in #1759
• Add encryption support using cryptography to Sessions implementation by @maxmekiska in #1674
• docs: reformat run module docstrings to Google style for better mkdocstrings rendering by @MuhammedSuhaib in #1740
• fix: #1668 Handoffs with gpt-5* model + store=False + remove_all_tools fails due to 404 error response by @ihower in #1757
• Add warning for agent names that transform into conflicting function names by @ihower in #1758
• feat: #1731 Enable developers to use Annotated types for function tool param description by @seratch in #1753
• feat: add more options to Agent#as_tool function by @seratch in #1751
• Export user_agent_override contextmanager by @jiwon-oai in #1768
• v0.3.1 by @rm-openai in #1773

New Contributors

• @MariumAfzal2703 made their first contribution in #1718
• @Wujiaxuan007 made their first contribution in #1703
• @LokinDev made their first contribution in #1739
• @AnkanMisra made their first contribution in #1745
• @serialx made their first contribution in #1744
• @notV3NOM made their first contribution in #1749
• @maxmekiska made their first contribution in #1674
• @jiwon-oai made their first contribution in #1768

Full Changelog: v0.3.0...v0.3.1