Simon Willison released LLM 0.32 with a redesigned logging system, new default models, and a major update to the llm-anthropic plugin.
Simon Willison released LLM 0.32 on 4 August 2026. The release adds visible reasoning traces, server-side provider tools, redesigned SQLite logs, and support for the OpenAI Responses API. The version also introduces new models, including GPT-5.6 Luna as the new default.
Simon Willison maintains LLM as a command-line tool and Python library. The release follows the initial launch of the project and includes a major update to the llm-anthropic plugin. The Anthropic plugin adds WebSearch, WebFetch, CodeExecution, and AnthropicMCP tools.
Builders can run one-off prompts against any OpenAI-compatible endpoint with the new llm openai endpoint command, which does not log them. The ability to hide reasoning traces with -R keeps piped output clean for automation. Server-side tools like CodeInterpreter and WebSearch now run as part of the model request, allowing tool use within a single API call.
Simon Willison called LLM 0.32 the most significant version since the project’s launch. The Python API changes and server-side tool support give builders new ways to work with model requests. Watch for the llm openai endpoint command and AnthropicMCP support as the project evolves.
What matters
- LLM 0.32 displays reasoning traces to standard error and supports server-side tools.
- Builders can now pipe reasoning traces separately from output and use provider tools in one command.
- Watch for the new llm openai endpoint command and the updated llm-anthropic plugin with AnthropicMCP support.
Why it matters
Watch for the new llm openai endpoint command and the updated llm-anthropic plugin with AnthropicMCP support.
This GenAI News article was prepared in original wording using reporting and materials published by Simon Willison’s Weblog. Source reference: https://simonwillison.net/2026/Aug/4/new-release-of-llm/#atom-everything.
Drafted by the GenAI News review pipeline.