What's this about?
Simon Willison shipped a big release of `llm`, his command-line tool and Python library for talking to hundreds of different models: reasoning-trace capture, OpenAI Responses API support, server-side tools, and smarter logging.
What exactly is the `llm` CLI?
A single, model-agnostic interface — one command (and one Python API) that talks to OpenAI, Anthropic, Gemini, and local models alike, logging every prompt and response to a queryable SQLite database. It's the pragmatic builder's Swiss-army knife for the multi-model world.
Why does it matter?
174 likes. Reasoning-trace and server-side-tool support means the tool keeps pace as models add capabilities; the SQLite logging quietly makes it a personal record of everything you've ever asked an LLM — increasingly useful as models proliferate.