Source · release

Pydantic AI v2.67.0

Author
Pydantic
Published
10 November 2026
Added
11 November 2026
Original
https://github.com/pydantic/pydantic-ai/releases/tag/v2.67.0

Summary

Pydantic AI 2.67.0 (2026-11-10) extends durable execution to sub-agents, makes the repository context loader pick up nested instruction files, and adds the newest models from Anthropic, Google, OpenAI, and Moonshot.

Key points

  • Durable sub-agents: SubAgents accepts dynamic_agents and task_control under Temporal and DBOS. With DBOS, tool calls can run as child workflows, in parallel when a workspace is attached. TrajectoryJudge evaluations are checkpointed when the judge has an id.
  • Nested instructions: RepoContext loads one instruction file per directory (with a walk_up option) and shows nested files when the model edits in that directory, capped by nested_max_chars. Workspace.authority_of sets per-path authority.
  • Models: Claude Fable 5.2 and Mythos 5.2 (also on Bedrock and OpenRouter), Gemini 4 Carbon on GoogleModel, gpt-image-2.6-flare and -sunburst, Kimi K3 on Bedrock; OpenAI fast/ultrafast service tiers; Anthropic web_search_20260318 and web_fetch_20260318 through provider_settings.
  • Other: dequeue and set_enqueue_priority change queued messages before delivery; KeenableSearch adds keyless web research.
  • Upgrade: pip install -U pydantic-ai.

Why it matters

Nested instruction loading mirrors how coding agents read per-folder CLAUDE.md or AGENTS.md files, and durable sub-agents let long multi-agent runs survive restarts.