Skip to content
Agent Radar

Pydantic AI 2.55.0 adds Conversation and prompt caching

Updated Pydantic AI

Summary

Pydantic AI 2.55.0, released October 9, 2026, requires Python 3.11 and adds Conversation, a unified cache setting and Postgres stores.

Tool
Pydantic AI
Category
Frameworks
Date
Open source
Yes
License
MIT
Models
OpenAIAnthropicGoogleclaude-haiku-5-5
Pricing
Not stated in the sources we cited.
Sources

Pydantic AI v2.55.0 was published on GitHub on October 9, 2026 by the pydantic/pydantic-ai project. According to the release notes, it raises the minimum Python version to 3.11 and adds a Conversation type, a unified prompt-caching setting and Postgres-backed stores.

Release facts

FactDetail
ProductPydantic AI (pydantic/pydantic-ai)
Versionv2.55.0
Release dateOctober 9, 2026
LicenseMIT, as reported by the GitHub API
Where to get itGitHub release tag v2.55.0 at https://github.com/pydantic/pydantic-ai/releases/tag/v2.55.0

What changed

  • Python 3.11 required. Every package now requires Python 3.11 or newer; Python 3.10 installs resolve to 2.54.0 and earlier.
  • Conversation. A new Conversation carries and stores a conversation between runs and is accepted as conversation= by every entry point.
  • Unified prompt caching. A cache setting and Caching capability cover cross-provider prompt caching. With cache=True on Anthropic's automatic caching, instructions and tool definitions are cached too, and harness Coder turns prompt caching on by default.
  • Cache diagnostics. Prompt cache diagnostics are recorded in provider_details (on by default for OpenAI Responses, anthropic_cache_diagnostics opt-in), and cache health is reported through pydantic_ai.cache.* span attributes.
  • Postgres stores. PostgresStepStore and PostgresMediaStore are added as hosted Postgres backends for messages and media.
  • Faster import. import pydantic_ai time is halved by loading pydantic_ai.mcp only when an MCP capability needs it.

The notes also list support for claude-haiku-5-5, new realtime voice options for Azure AI Voice Live and GPT-Live, and an OpenAIDecisionsModel backend with image input.

Questions

Which Python versions does 2.55.0 support? The release notes say every package requires Python 3.11 or newer, and Python 3.10 installs resolve to 2.54.0 and earlier.

Does this release change durable execution behavior? Yes. The release notes list changes that record tool requests of ExaSearch, YouSearch, LocalStack and other harness capabilities under durable execution, and keep default run_id and conversation_id stable when a durable run is re-executed.

Is Pydantic AI open source? The GitHub repository reports an MIT license. The product site also labels Pydantic Evals as open source.

See the pydantic-ai hub for more, browse the Frameworks category, or compare with the adk-python 2.12.0 launch.

Alternatives

Pydantic AI compared with its alternatives
ToolLicensePricing note
Pydantic AIThis launchPydantic AI is a model-agnostic Python framework for building typed AI agents and applications.
MITNot stated
LangGraphLangGraph is an open-source orchestration runtime for building stateful, resilient agents.
MITNot stated
MastraTypeScript agent framework. Mastra Code is the terminal agent; AgentController is the extracted loop, in @mastra/core 1.47.0 or later.
Open sourceMastra Code and AgentController are announced as open source. No license fee is stated in the posts we cited.
AgnoAgno is an orchestration and service layer for agents, exposing them through a REST API and an MCP server.
Apache-2.0Not stated
adk-pythonAn open-source, code-first Python toolkit for building, evaluating and deploying AI agents.
Apache-2.0Not stated

Questions

What is Pydantic AI?

The product site describes it as a model-agnostic Python framework featuring type safety, structured outputs, validation and observability.

Is Pydantic AI open source?

The GitHub repository reports an MIT license.

Which models does it support?

The product site says it supports OpenAI, Anthropic, Google and more.

Filed on Agent Radar. Claims in this note are limited to the sources above.