v1.5.0 (2025-10-24)

What's Changed

• Introduce new graph API in beta by @dmontagu in #2982
• Preformat run graph/node span names for other OTel backends by @alexmojaki in #3244
• Ensure that google-genai doesn't close httpx client provided by Pydantic AI or user by @DouweM in #3243

Full Changelog: v1.4.0...v1.5.0

v1.4.0 (2025-10-23)

What's Changed

• Support OpenAI and Anthropic native MCP support via MCPServerTool builtin tool by @Artui in #3101, #3239
• Use correct agent's instructions when telling model to retry output by @dsfaccini in #3209
• Raise clear error when any Google content filter is hit resulting in empty response by @DouweM in #3236
• Expand docs for pydantic-evals by @dmontagu in #3213

New Contributors

• @Artui made their first contribution in #3101

Full Changelog: v1.3.0...v1.4.0

v1.3.0 (2025-10-22)

What's Changed

• Raise IncompleteToolCall when token limit is reached during generation of tool call by @erhuve in #3137
• Include evals report averages in span attributes by @cetra3 in #3053
• Make AbstractBuiltinTool serializable and work with durable execution by @DouweM in #3176
• Ignore empty text deltas when streaming gpt-oss on Bedrock by @DouweM in #3215
• Ignore empty text deltas when streaming gpt-oss via Ollama by @DouweM in #3216
• docs: Replace deprecated Cohere model 'command' with 'command-r7b-12-… by @sijanonly in #3201
• Include all usage fields in OTel attributes by @DouweM in #3221
• Correct Prefect .serve example by @desertaxle in #3192
• Ensure toolset spans (e.g. MCP sampling) are nested under agent run span by @DouweM in #3224
• Update genai-prices, fix default api_flavor, fixes anthropic usage extraction by @alexmojaki in #3226
• feat(gateway): support AWS Bedrock by @Kludex in #3203
• Add OVHcloud AI Endpoints provider by @eliasto in #3188
• Use gateway/<upstream_provider>: as provider name prefix for Gateway by @DouweM in #3229
• Add http_client option to GoogleProvider, support api_key for Vertex AI, and use Pydantic AI's cached httpx client by default by @DouweM in #3217

New Contributors

• @erhuve made their first contribution in #3137
• @cetra3 made their first contribution in #3053
• @sijanonly made their first contribution in #3201
• @eliasto made their first contribution in #3188

Full Changelog: v1.2.0...v1.3.0

v1.2.0 (2025-10-20)

What's Changed

• Strip markdown fences when generating evals datasets by @jlegewie in #3141
• Include final_result agent span attribute after streaming by @DouweM in #3170
• Add Claude Haiku 4.5 model by @rian-dolphin in #3183
• change paig base url by @samuelcolvin in #3200
• Extract openai usage using genai-prices by @alexmojaki in #3123

New Contributors

• @jlegewie made their first contribution in #3141
• @rian-dolphin made their first contribution in #3183

Full Changelog: v1.1.0...v1.2.0

v1.1.0 (2025-10-15)

Features in v1.1.0

• Add support for durable execution with Prefect by @desertaxle in #3074, #3156

Selected Features since v1.0.0

• Support image generation and output with Google and OpenAI by @DouweM in #2970
• Add Pydantic AI Gateway provider by @Kludex in #2816, #2863
• Add support for previous_response_id from OpenAI Responses API by @GDaamn in #2756
• Add support for durable execution with DBOS by @qianl15 in #2638
• Built-in tool call streaming from OpenAI, Google, Anthropic by @DouweM in #2877
• Support Anthropic built-in memory tool by @DouweM in #3042
• Support text, JSON, XML and YAML DocumentUrl and BinaryContent on OpenAI by @pulphix in #2851
• Added MCP metadata and annotations to ToolDefinition.metadata for use in filtering by @ChuckJonas in #2880
• Let agent name be overridden contextually by @MinuraPunchihewa in #3094
• Support contextually overriding agent instructions by @mwildehahn in #2926
• Tools can now return AG-UI events separate from result sent to model by @DouweM in #2922
• Add AgentRunResult.response convenience method to get latest model response by @DouweM in #2970
• Add ModelResponse.text, thinking, files, images, tool_calls, and builtin_tool_calls convenience methods by @DouweM in #2970
• Add Agent.run_stream_events() convenience method wrapping run(event_stream_handler=...) by @DouweM in #3084

Other Changes in v1.1.0

• Record instructions on the agent run span even when they are dynamic by @dmontagu in #3131
• Add description arg to tool function decorators by @zhcn000000 in #3153
• fix(gateway): update base_url by @Kludex in #3164
• Document OpenAI-compatible provider prefixes by @DouweM in #3169
• Explicitly request image response modality from Google API when model supports it by @DouweM in #3172

New Contributors

• @zhcn000000 made their first contribution in #3153
• @certainly-param made their first contribution in #3133

Full Changelog: v1.0.18...v1.1.0

v1.0.18 (2025-10-13)

What's Changed

• Add render method to EvaluationReport class by @dmontagu in #3116
• Omit previous_response_id when unset instead of sending null to OpenAI Responses by @DouweM in #3134
• Include all API docs in llms.txt by @DouweM in #3152
• Add anyio and httpcore to Temporal passthrough modules by @slumbi in #3147
• Add Nebius AI Studio provider support by @antoncp in #3124
• Add new ToolCallPart.id field for OpenAI Responses by @DouweM in #3151

New Contributors

• @slumbi made their first contribution in #3147
• @antoncp made their first contribution in #3124
• @desertaxle made their first contribution in #3074

Full Changelog: v1.0.17...v1.0.18

v1.0.17 (2025-10-09)

What's Changed

• Let builtin_tools be specified at agent run time by @safina57 in #3009
• Make BinaryImage work with inline-snapshot by @alexmojaki in #3125
• Document InstrumentationSettings(version=3) by @alexmojaki in #3127

Full Changelog: v1.0.16...v1.0.17

v1.0.16 (2025-10-08)

What's Changed

• Add support for datetime.time/timedelta to format_as_xml by @nurikk in #3087
• Ensure graph persistence snapshots are not mutated when run is resumed by @lienminhquang in #3077
• Use Sequence[ModelMessage] instead of list for method arg types by @moritzwilksch in #3040
• Let agent name be overridden contextually by @MinuraPunchihewa in #3094
• Respect FileUrl.force_download flag in OpenAI Chat and Responses models by @nicolas-chaulet in #3106
• Ensure FileUrl and BinaryContent without identifier are valid by @DouweM in #3110
• Don't include ToolResultPart for external tool call when streaming by @DouweM in #3112
• Fix token usage for anthropic streaming by @alexmojaki in #3111 and #3115
• plain text docs, fix redirects and add logfire by @samuelcolvin in #3096

New Contributors

• @nurikk made their first contribution in #3087
• @lienminhquang made their first contribution in #3077
• @MinuraPunchihewa made their first contribution in #3094
• @nicolas-chaulet made their first contribution in #3106

Full Changelog: v1.0.15...v1.0.16

v1.0.15 (2025-10-03)

What's Changed

• Support image generation and output with Google and OpenAI by @DouweM in #2970
• Add AgentRunResult.response convenience method to get latest model response by @DouweM in #2970
• Add ModelResponse.text, thinking, files, images, tool_calls, and builtin_tool_calls convenience methods by @DouweM in #2970
• Add Agent.run_stream_events() convenience method wrapping run(event_stream_handler=...) by @DouweM in #3084
• Add content (e.g. files) returned by tool to FunctionToolResultEvent by @DouweM in #3082
• Set MCPServer id and tool_prefix in load_mcp_servers by @DouweM in #3052
• Add latest gemini 2.5 flash(-lite) model names and aliases by @moritzwilksch in #3060
• Support enums in format_as_xml by @DouweM in #3064
• Fix dataset serialization when inputs have discriminators with defaults by @DouweM in #3079
• Fix parallel tool call limit enforcement by @tradeqvest in #2978

Full Changelog: v1.0.14...v1.0.15

v1.0.14 (2025-10-02)

What's Changed

• Remove leftover debug print statement by @KostyaGukish in #3070
• Fix duplicate output tool return part when concatenating first run messages with follow-up new_messages by @DouweM in #3075

New Contributors

• @KostyaGukish made their first contribution in #3070

Full Changelog: v1.0.13...v1.0.14