LLM
Open-source CLI and Python library for talking to dozens of LLMs from the terminal.
About
LLM by Simon Willison is among the most widely used open-source tools for working with language models from the command line, with over 11,000 GitHub stars. It provides a unified interface for prompting remote APIs (OpenAI, Anthropic, Gemini) and local models, with a plugin architecture that adds new model providers rather than baking them in.
Every prompt and response is logged to a local SQLite database by default, and the logs can be browsed with Datasette. Recent versions add tool/function calling from the CLI, structured output via JSON schemas, embeddings with semantic search for lightweight command-line RAG, and reusable prompt templates.
What it can do
- Run prompts against dozens of remote and local models from one CLI
- SQLite logging of every prompt and response
- Plugin ecosystem for models, embeddings, and custom functions
- Embeddings, structured output, and tool calling from the terminal
Setup
pip install llm (or brew install llm), then configure API keys with llm keys set openai.
Prerequisites
Python 3.8+ and API keys for whichever model providers you want to use.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 106; Trending score 90/100 from researcher assessment.
★ 11,508 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.