LLM CLI: llm#

llm is a command-line utility for interacting with large language models—simplifying prompts, managing models and plugins, logging every conversation, and extracting structured data for pipelines.

Language models on the command-line w/ Simon Willison

Basic Usage#

Install llm. Then set up your OPENAI_API_KEY environment variable. See Getting started.

TDS Students: See Large Language Models for instructions on how to get and use OPENAI_API_KEY.

# Run a simple prompt
llm 'five great names for a pet pelican'

# Continue a conversation
llm -c 'now do walruses'

# Start a memory-aware chat session
llm chat

# Specify a model
llm -m gpt-4.1-nano 'Summarize tomorrow’s meeting agenda'

# Extract JSON output
llm 'List the top 5 Python viz libraries with descriptions' \
  --schema-multi 'name,description'

Or use llm without installation using uvx:

# Run llm via uvx without any prior installation
uvx llm 'Translate "Hello, world" into Japanese'

# Specify a model
uvx llm -m gpt-4.1-nano 'Draft a 200-word blog post on data ethics'

# Use structured JSON output
uvx llm 'List the top 5 programming languages in 2025 with their release years' \
  --schema-multi 'rank,language,release_year'

Key Features#

  • Interactive prompts: llm '…' — Fast shell access to any LLM.
  • Conversational flow: -c '…' — Continue context across prompts.
  • Model switching: -m MODEL — Use OpenAI, Anthropic, local models, and more.
  • Structured output: llm json — Produce JSON for automation.
  • Logging & history: llm logs path — Persist every prompt/response in SQLite.
  • Web UI: datasette "$(llm logs path)" — Browse your entire history with Datasette.
  • Persistent chat: llm chat — Keep the model in memory across multiple interactions.
  • Plugin ecosystem: llm install PLUGIN — Add support for new models, data sources, or workflows. (Language models on the command-line - Simon Willison’s Weblog)

Practical Uses#

Backward Devtools Git Forward