v1.22.0 (2025-11-21)

What's Changed

• Add OpenRouterModel as OpenAIChatModel subclass with additional feature support by @ajac-zero in #3089
• Make FallbackModel fall back on all model API errors, not just HTTP status 400+ by @dsfaccini in #3494
• Don't skip model request when history ends on response but there are new instructions by @DouweM in #3509
• Revert "feat: enforce message history starts with user message" by @DouweM in #3508
• Don't send Google messages with 0 parts by @DouweM in #3511
• Immediately raise error when response is empty because of token limit by @DouweM in #3512
• Relax UserError into a warning when state deps is not provided with AG-UI by @MikeRyanDev in #3510
• Add type stubs for some third-party libraries by @lars20070 in #3443
• Add document to allowed cacheable_types for anthropic by @marpulli in #3513
• Don't insert empty ThinkingPart when Google response ends in text with thought_signature by @DouweM in #3516
• Fix deprecated GeminiModel structured output and tool call thought signatures by @DouweM in #3517

New Contributors

• @marpulli made their first contribution in #3513

Full Changelog: v1.21.0...v1.22.0

v1.21.0 (2025-11-20)

What's Changed

• Feature: MCP client Resources support by @fennb in #3024
• Expose MCP server instructions in MCPServer.instructions property by @mochow13 in #3431
• Add BinaryContent.from_path convenience method by @dsfaccini in #3482
• Enforce message history starts with user message by @Pavanmanikanta98 in #3440
• Always strip Markdown fences from structured output by @DouweM in #3475
• Fix Gemini nested tool argument schemas by removing title from $defs by @conradlee in #3487
• Fix issue with Gemini and Prefect by removing unused GoogleModel._url by @DouweM in #3499

New Contributors

• @Pavanmanikanta98 made their first contribution in #3440
• @fennb made their first contribution in #3024

Full Changelog: v1.20.0...v1.21.0

v1.20.0 (2025-11-18)

What's Changed

• Add support for Gemini 3 Pro by @DouweM in #3464
• Support Gemini enhanced JSON Schema features by @conradlee in #3357
• Wrap GoogleModel google.genai.errors.APIError in ModelHTTPError so it works with FallbackModel by @timothy-jeong in #3139
• fix: Change handling of empty state objects for AG-UI by @MikeRyanDev in #3462
• Adds {ModelRequest,ModelResponse}.metadata fields by @adtyavrdhn in #3422
• Make RunContext.usage available with Temporal by @mattbrandman in #3434
• Add ttl to CachePoint and Anthropic caching model settings by @Wh1isper in #3450
• Extract google model usage using genai-prices by @alexmojaki in #3466
• Update has_content method to check content data by @dsfaccini in #3436

New Contributors

• @conradlee made their first contribution in #3357
• @MikeRyanDev made their first contribution in #3462
• @timothy-jeong made their first contribution in #3139

Full Changelog: v1.19.0...v1.20.0

v1.19.0 (2025-11-17)

What's Changed

• Let metadata be passed to CallDeferred and ApprovalRequired exceptions and end up on DeferredToolRequests by @cjohnhanson in #3345
• Add AnthropicModel count_tokens and support UsageLimits.count_tokens_before_request by @nathan-gage in #3458
• Fix bug with running graphs in temporal workflows by @dmontagu in #3460
• Fix Gateway with Temporal by adding sniffio to sandbox passthrough modules by @DouweM in #3455
• Fix Gateway links in code blocks by @NicolasPllr1 in #3444

New Contributors

• @NicolasPllr1 made their first contribution in #3444

Full Changelog: v1.18.0...v1.19.0

v1.18.0 (2025-11-14)

What's Changed

• Add Anthropic prompt caching support by @ronakrm in #3363
• Bump openai to v2.8.0 (v1 still supported), add GPT-5.1 to known model names by @DouweM in #3419
• Bump google-genai to v1.50.1 and remove patches by @DouweM in #3340
• Bump temporalio to v1.19.0 and use SimplePlugin by @DouweM in #3214

New Contributors

• @notrudyyy made their first contribution in #3421
• @ronakrm made their first contribution in #3363

Full Changelog: v1.17.0...v1.18.0

v1.17.0 (2025-11-13)

What's Changed

• Add Pydantic AI Gateway docs by @laisbsc in #3402
• Make FastMCPToolset work with Temporal by @DouweM in #3413
• Support mcp.json environment variable expansion in load_mcp_servers() by @wgillett in #3380

New Contributors

• @laisbsc made their first contribution in #3402
• @wgillett made their first contribution in #3380

Full Changelog: v1.16.0...v1.17.0

v1.16.0 (2025-11-13)

What's Changed

• refactor(gateway): add upstream_provider back by @Kludex in #3391

Full Changelog: v1.15.0...v1.16.0

v1.15.0 (2025-11-12)

What's Changed

• Wrap BedrockConverseModel errors in ModelHTTPError to work well with FallbackModel by @cjermain in #3377
• Add unique run_id to run, run result, and message (request, response) classes by @adtyavrdhn in #3366
• Add BedrockConverseModel.count_tokens so it works with UsageLimits.count_tokens_before_request by @DenysMoskalenko in #3367

Full Changelog: v1.14.1...v1.15.0

v1.14.1 (2025-11-11)

What's Changed

• Fix Vercel AI tool input/output always showing up as a JSON string rather than object by @DouweM in #3399
• Make ModelRetry hashable by @nathan-gage in #3394
• Don't run CI steps that require secrets on PRs from fork branch to fork main by @lars20070 in #3385

New Contributors

• @lars20070 made their first contribution in #3385
• @nathan-gage made their first contribution in #3394

Full Changelog: v1.14.0...v1.15.0

v1.14.0 (2025-11-10)

What's Changed

• Allow custom provider factory to be passed into infer_model by @slkoo-cc in #3341
• Fix error when Google returns only empty text parts by @naveen-corpusant in #3388
• Docs: AWS Bedrock retry behavior by @cjermain in #3379
• docs: fix link to AG-UI Dojo by @echarles in #3387

New Contributors

• @cjermain made their first contribution in #3379
• @echarles made their first contribution in #3387

Full Changelog: v1.13.0...v1.13.1