What changed
v2.66.0 โ v2.67.0: adds Fable 5.2 / Mythos 5.2, Gemini 4 Carbon, Kimi K3 and Kimi K3.1 on Bedrock, OpenAI gpt-image-2.6-flare and gpt-image-2.6-sunburst; ships new KeenableSearch capability (keyless web research), DBOS child workflow tool calls, SubAgents durable execution fix for dynamic_agents, and queue control (dequeue, set_enqueue_priority). Also includes 20+ compatibility notes (three hard breaks in the Breaking Changes section above).
TL;DR
PydanticAI v2.67.0 adds claude-fable-5-2, claude-mythos-5-2, gemini-4-carbon, kimi-k3, and OpenAI image models in a single release, 36 hours after the Fable 5.2 launch; also ships KeenableSearch for keyless web research and DBOS child workflow support.
Developer signal
pip install --upgrade pydantic-ai to get v2.67.0. To use Fable 5.2: model='claude-fable-5-2' or model='claude-mythos-5-2' โ both work on direct Anthropic, Bedrock Mantle (gateway/bedrock:global.anthropic.claude-fable-5-2), and OpenRouter. The effort parameter controls thinking depth (replaces budget_tokens โ Fable 5.2 follows the Fable 5.2 API conventions covered in the Nov 9 digest). For KeenableSearch: add KeenableSearch() to your agent's capabilities list โ it performs web research without requiring an API key, using Claude's built-in search tools. For DBOS child workflows: add child_workflow=True to the DBOSDurability metadata on a tool call; child-workflow tools in the same workspace run in parallel automatically. Read the full compatibility notes before upgrading if you use: SubAgents with dynamic_agents, RepoContext, OpenAI continuation chains, CapabilityCreation, CodeMode, or provider-specific native-tool settings.