What's Changed

• Allow TemporalAgent to switch model at agent.run-time by @mattbrandman in #3537
• Add DynamicToolset support in Temporal by @dsfaccini in #3682
• Add support for ImageGenerationTool output_compression and output_format for Vertex AI Gemini image models by @h0rv in #3759
• Update known Groq model names (add production/preview, remove deprecated) by @hima12-awny in #3774
• Add model profile flag for APIs that support native output but still require JSON schema in instructions by @xcpky in #3744
• Set ToolRetryError message by @jamesaud in #3718
• Fix StreamedRunResult.get_output() creating duplicate messages if stream_output() has already been called by @DouweM in #3777
• Re-add clai --help output to clai README by @dsfaccini in #3766

New Contributors

• @hima12-awny made their first contribution in #3774
• @jamesaud made their first contribution in #3718

Full Changelog: v1.36.0...v1.37.0

What's Changed

• Add AI SDK data chunk ID and tool approval types by @bendrucker in #3760
• Ensure type field when converting const to enum in GoogleJsonSchemaTransformer by @majiayu000 in #3751
• Bump google-genai to 1.56.0 by @petersli in #3770

New Contributors

• @majiayu000 made their first contribution in #3751
• @bendrucker made their first contribution in #3760

Full Changelog: v1.35.0...v1.36.0

What's Changed

• Add FileSearchTool with support for OpenAI and Google by @gorkachea in #3396
• Add support for ImageGenerationTool.size to Gemini image models by @h0rv in #3720
• Adding Gemini 3 flash by @jhammarstedt in #3755
• Add Alibaba Cloud DashScopeProvider and support audio input for Qwen Omni by @Pavanmanikanta98 in #3596
• Added support for AG-UI Multi-modal Messages by @bojan2501 in #3715
• Set timestamps on AG-UI events by @thejens in #3742
• Support OpenAI reasoning summary option 'auto' by @DouweM in #3753
• Operate on a deepcopy of $defs in JsonSchemaTransformer instead of the original schema by @thomasst in #3758
• Fix typing issue when using UIAdapter.dispatch_request with agent with output_type by @DouweM in #3749

New Contributors

• @h0rv made their first contribution in #3720
• @thejens made their first contribution in #3742
• @gorkachea made their first contribution in #3396
• @bojan2501 made their first contribution in #3715
• @thomasst made their first contribution in #3758

Full Changelog: v1.34.0...v1.35.0

What's Changed

• Add Web Chat UI for any agent that can be launched using clai web or Agent.to_web() by @dsfaccini in #3456
• Support FileUrl.force_download in AnthropicModel and OpenAIResponsesModel by @dsfaccini in #3694
• Fix using sync history processors, instructions functions, and output functions with TemporalAgent by @dsfaccini in #3704
• Make OpenRouterProvider and DeepSeekProvider __init__ overloads less restrictive by @xcpky in #3739
• Bump min version of griffe to 1.14.0 by @jerry-reevo in #3743

Full Changelog: v1.33.0...v1.34.0

What's Changed

• Pass s3:// file URLs directly to API in BedrockConverseModel by @mochow13 in #3663
• Insert agent instructions after system_prompts for models that don't natively support instructions by @siddhantbhagat8 in #3614
• Bump google-genai to 1.55 by @dsfaccini in #3727
• docs: Update mkdocstrings-python to 2.x and fix cross-reference paths by @dsfaccini in #3706

New Contributors

• @siddhantbhagat8 made their first contribution in #3614

Full Changelog: v1.32.0...v1.33.0

What's Changed

• Add tool timeout support by @DEENUU1 in #3594
• Let TemporalAgents be registered to a Temporal workflow using __pydantic_ai_agents__ field by @adtyavrdhn in #3676
• Make end_strategy also work for output tools, not just tools by @Danipulok in #3523
• Replace OTel events with logs by @alexmojaki in #3641
• Fix UIAdapter.dispatch_request typing by @dmontagu in #3721

New Contributors

• @DEENUU1 made their first contribution in #3594

Full Changelog: v1.31.0...v1.32.0

What's Changed

• Add prompt caching support for AWS Bedrock by @DenysMoskalenko in #3438
• Add Agent.output_json_schema() method by @g-eoj in #3454
• Allow custom clientInfo when connecting to MCP servers by @atinylittleshell in #3572
• Add provider_url to ModelResponse and use it in cost() by @mahiro72 in #3648
• Add GPT-5.2 and bump openai to v2.11.0 by @DouweM in #3713
• Allow model to be a string in LLMJudge by @samuelcolvin in #3711
• Fix OpenAIResponsesModel web search find_in_page action handling by @DouweM in #3709

New Contributors

• @mahiro72 made their first contribution in #3648
• @atinylittleshell made their first contribution in #3572

Full Changelog: v1.30.0...v1.31.0

What's Changed

• Restore compatibility with openai v1 by @DouweM in #3707
• Temporarily revert support for OpenAI prompt caching options (#3678) by @DouweM in #3707
  • This feature will return in v1.31.0

Full Changelog: v1.30.0...v1.30.1

What's Changed

• Add prompt caching options to OpenAIChatModelSettings by @kichanyurd in #3678
• Support multi-modal output in LLMJudge by @Clement-Lelievre in #3696
• Add CerebrasModel by @sebastiand-cerebras in #3486
• Fix GraphBuildingError when using g.stream multiple times by @adtyavrdhn in #3695
• Fix BedrockConverseModel error when ModelResponse.parts is empty by @ffineis in #3689
• Pin deno to v2.5.x by @DouweM in #3700

New Contributors

• @ffineis made their first contribution in #3689
• @sebastiand-cerebras made their first contribution in #3486
• @kichanyurd made their first contribution in #3678
• @Clement-Lelievre made their first contribution in #3696

Full Changelog: v1.29.0...v1.30.0

What's Changed

• Add support for aspect ratio in Gemini image generation by @mwildehahn and @ajac-zero in #3672
• Pass container_id back to Anthropic API by @mattbrandman in #3637
• Upgrade temporalio to 1.20.0 by @akrivka in #3679
• Fix error when OpenRouter response includes file annotations by @dsfaccini in #3671
• Suppress broken resource errors if cancelling by @dmontagu in #3675
• Fix GoogleModel thinking signature not stored on tool calls when streaming by @jerry-heygen in #3647
• Don't require anthropic dependency when using Anthropic model with other provider by @dsfaccini in #3652

New Contributors

• @akrivka made their first contribution in #3679

Full Changelog: v1.28.0...v1.29.0