Pydantic AI 2.56.0 released
A feature release adding browser execution on host models, Gemini 4 Argon support, Render Workflows integration, and numerous bug fixes.
What changed in 2.56.0
- BrowserUse now runs on the host model by default; browser-use telemetry requires opt-in.
- Skills layer can project personal libraries, name skill directories, and render skills for host commands.
- Added SmolSandbox workspace for running harness capabilities in local microVMs.
- Support for Gemini 4 Argon model on GoogleModel with extended output decoding and 20-minute default timeout.
- Multiple fixes for tool handling, native tool events, Bedrock rejections, and memory snapshots across various model providers.
Check before you update
- BrowserUse runs on host model by default instead of previous behavior; browser-use telemetry is now opt-in.
- WarnNearLimits removed from harness Coder capability.
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 | 13 stable releases in the last 30 days; 2.56.0 came 4 days 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 3.9% |
| Gid Score | 52 #15 in Agent frameworks and SDKs |