New release of LLM adds support for reasoning traces, OpenAI Responses, server-side tools, and smarter logging

Simon Willison released LLM zero point thirty-two this morning—the most significant update to his command-line tool for working with language models since the project launched. The new version displays reasoning traces from reasoning models, adds server-side tools like code execution and web search from OpenAI and Anthropic, and uses the inexpensive GPT-five point six Luna model as its default. The tool is increasingly agent-like, now able to loop through multiple tool calls, pause for human approval, and resume from saved conversation history. Willison is considering formalizing these capabilities as core agent abstractions in future versions of the library.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton