Type: url
Original: https://simonwillison.net/2025/May/27/llm-tools/
Status: generated
Language: en
Summary
Simon Willison announces LLM 0.26, which adds tool support to his LLM command-line tool and Python library so that models from OpenAI, Anthropic, Gemini and local Ollama models can call tools written as Python functions. The post walks through built-in demo tools, tool plugins such as llm-tools-simpleeval and llm-tools-datasette, ad-hoc tools passed with —functions, and the new model.chain() Python API. It also explains why tool support took so long — waiting for vendors to converge on the same underlying pattern — describes ‘agents’ as converging on ‘tools in a loop’, and names Model Context Protocol client support as a next step.