Pydantic AI now turns on prompt caching with one capability.
Anthropic and Bedrock cache nothing unless the request asks. The Caching capability asks for you: tool definitions, instructions and the conversation, one setting instead of each provider's.
https://t.co/LXH3aKvreE
Added @a1zhang 's Speculative Programmatic Tool Calling to @pydantic AI, and I've been getting nice results from it.
There is pretty much no drawback to Code Mode that I can see.
https://t.co/lIq3U1V2CP has had it for awhile now, and I added a little statusline to show you how much time you save from speculated tool calls.
It's basically just free speed.
https://t.co/L13luftAz4
Imagine being an unnamed Fortune 2 company with 100k people using a @Pydantic AI-based coding agent to streamline their jobs.
What if Pydantic Logfire let you observe entire fleets of agents, centrally manage skills, MCP, guardrails & memory, and auto-suggest ones from traces?
@seanmonstar I've been slowly working on multipart/form-data support in this issue, but I don't really know the appropriate way to integrate this into warp: https://t.co/u4Bvh5jTZO