Pydantic AI 2.63.0 released
A feature release adding new capabilities, models, and code execution modes while fixing compatibility issues and bugs across multiple providers.
What changed in 2.63.0
- Added IPythonCodeMode for code execution, DetectLoops capability to prevent repeated actions, and PostgresSpendStore for shared budget management.
- Introduced gpt-6.1-terra model support and ag-ui-protocol 1.0, plus task_timeout for Dataset.evaluate.
- Changed CodeMode default to dynamic_catalog=True so revealed tools append instead of rewriting cached prompts, and LLMReminder default interval to 10.
- Fixed tag escaping in system messages, Azure rate-limit handling, batch request withdrawal, and multiple provider-specific issues.
- Improved history replay for encrypted data, deferred capability loading, and UI streaming for thinking changes and transcripts.
Check before you update
- CodeMode now defaults to dynamic_catalog=True, changing how revealed tools are handled in cached prompts.
- Gemini Live reconnect replay now passed to GoogleRealtimeConnection.dial as keyword-only replay argument.
- LLMReminder in Harness now defaults to interval=10 instead of generating on every request.
Summarised by AgentGid from Pydantic's official release notes on GitHub, as reported there; read the original before relying on it. Price, status and release dates come from our daily data.
Pydantic AI at a glance
| What it is | A Python agent framework from the Pydantic team, built around type-safe structured outputs, dependency injection and model-agnostic tool calling. |
|---|---|
| Latest version | 2.71.0, Nov 19, 2026. This page is about an earlier release. |
| Release pace | 15 stable releases in the last 30 days; 2.63.0 came 1 day after the previous one |
| Price | Free (OSS). Pydantic AI is open source and free to use; you pay only for the models and infrastructure it runs on. (checked Nov 4, 2026) |
| Downloads, last 7 days | 1.6M up 1.1% |
| Gid Score | 53 #15 in Agent frameworks and SDKs |