v0.2.7 (2025-05-24)

What's Changed

• Support streaming tool calls from models that pass args as None when there are no function parameters by @DouweM in #1802
• Stream tool calls and structured output from Anthropic as it's received instead of in one go by @DouweM in #1669
• Remove hardcoded n parameter from OpenAIModel requests by @kiqaps in #1807
• Add tool_prefix option to MCP servers and error on conflicting tool names by @Wh1isper in #1266
• Allow RunContext to not be documented when require_parameter_descriptions=True as it's not passed to the model anyway by @Kevsnz in #1750
• Clean up dataclasses reprs by @dmontagu in #1812
• Addition: Make prog_name customizable by @fswair in #1804
• Use AsyncBeta instead of AsyncAnthropic by @Kludex in #1818

New Contributors

• @kiqaps made their first contribution in #1807
• @Kevsnz made their first contribution in #1750
• @fswair made their first contribution in #1804

Full Changelog: v0.2.6...v0.2.7

v0.2.6 (2025-05-20)

What's Changed

• Add prepare_tools param to Agent class by @rmaceissoft in #1474
• fix: remove forgotten print on GoogleModel by @Kludex in #1780
• fasta2a: be more strict on agent card by @Kludex in #1781
• fix: rename prepare_tools to _prepare_tools by @Kludex in #1789
• fix: create prompt history file when using Agent.to_cli() by @Kludex in #1791
• fix: add 'openrouter' string to OpenAIModel provider param by @Kludex in #1792

New Contributors

• @rmaceissoft made their first contribution in #1474

Full Changelog: v0.2.5...v0.2.6

v0.2.5 (2025-05-20)

What's Changed

• fasta2a: make capabilities required on AgentCard by @Kludex in #1733
• Fix formatting of ints in eval reports by @dmontagu in #1738
• Bugfix: record instructions properly on agent run span when using structured output by @dmontagu in #1740
• Add ability to specify the evaluation name for all provided Evaluators by @dmontagu in #1725
• Add include_binary_content flag to InstrumentationSettings, rename OTel attribute key from content to binary_content for BinaryParts by @alexmojaki in #1739
• Add logprobs to OpenAI model settings and response by @eliassoares in #1238
• Added vendor_id to the model response. by @peterHoburg in #1547
• Fix ImportError when opentelemetry is not installed by @dmontagu in #1773
• Add licenses to all packages. by @tobiasraabe in #1775
• Add OpenRouter provider by @Kludex in #1778
• Add Google GenAI provider by @Kludex in #1751

New Contributors

• @eliassoares made their first contribution in #1238
• @peterHoburg made their first contribution in #1547
• @iandolge made their first contribution in #1762
• @tobiasraabe made their first contribution in #1775

Full Changelog: v0.2.4...v0.2.5

v0.2.4 (2025-05-14)

What's Changed

• clai should not require OPENAI_API_KEY by @samuelcolvin in #1724

New Contributors

• @chandy made their first contribution in #1727

Full Changelog: v0.2.3...v0.2.4

v0.2.3 (2025-05-13)

What's Changed

• cli: add current path to sys.path by @Kludex in #1703
• evals: use the correct render object on include_metadata and include_expected_output by @lionpeloux in #1700
• Add A2A server by @Kludex in #1537
• Allow definition of Model Settings for LLMJudge by @assadyousuf in #1662
• improve messages coverage by @samuelcolvin in #1704
• Add direct public API by @samuelcolvin in #1599
• Ensure CLI prompt history directory and file exist before usage by @zboyles in #1707

New Contributors

• @lionpeloux made their first contribution in #1700
• @assadyousuf made their first contribution in #1662
• @zboyles made their first contribution in #1707

Full Changelog: v0.2.2...v0.2.3

v0.2.2 (2025-05-13)

What's Changed

• Add to_cli() method to Agent by @AndrewHannigan in #1642

New Contributors

• @AndrewHannigan made their first contribution in #1642

Full Changelog: v0.2.1...v0.2.2

v0.2.1 (2025-05-13)

What's Changed

• Add AWS Profile Feature by @prescod in #1634
• Change Agent.is_?_node() to use TypeIs instead by @chasewalden in #1622
• Add newline separation between Anthropic system prompts by @timesler in #1638
• cli: store prompt and config under ~/.pydantic-ai by @Kludex in #1701
• otel: send BinaryContent information by @Kludex in #1603

New Contributors

• @prescod made their first contribution in #1634
• @chasewalden made their first contribution in #1622

Full Changelog: v0.2.0...v0.2.1

v0.2.0 (2025-05-12)

Breaking Changes

See #1647 — usage makes sense as part of ModelResponse, and could be really useful in "messages" (really a sequence of requests and response). In this PR:

• Adds usage to ModelResponse (field has a default factory of Usage() so it'll work to load data that doesn't have usage)
• changes the return type of Model.request to just ModelResponse instead of tuple[ModelResponse, Usage]

Other Changes

• Add support for non-string enums in Gemini by @zaidhaan in #1564
• Fix parallel tool calls on Bedrock with Nova and Claude models by @pandersen-uncharted in #1656
• Fix Anthropic streaming by @samuelcolvin in #1686
• Fix serializability of ModelRequestParameters by @dmontagu in #1690
• Stop incorrectly treating side-by-side JSON schema $refs as recursion by @DouweM in #1697
• Set ToolCallPartDelta.tool_call_id when known from previous delta by @DouweM in #1694

New Contributors

• @zaidhaan made their first contribution in #1564

Full Changelog: v0.1.11...v0.2.0

v0.1.11 (2025-05-10)

What's Changed

• Switch CLI to clai by @samuelcolvin in #1504

Full Changelog: v0.1.10...v0.1.11

v0.1.10 (2025-05-06)

What's Changed

• Allow setting temperature to 0 on BedrockConverseModel by @pandersen-uncharted in #1631
• Add extra_headers to ModelSettings by @ramanan-ravi in #1644
• Add thinking_config to GeminiModel by @Kludex in #1609

New Contributors

• @pandersen-uncharted made their first contribution in #1631
• @ramanan-ravi made their first contribution in #1644

Full Changelog: v0.1.9...v0.1.10