llm 0.33

Simon Willison's Weblog notes that version zero point three three of the llm command‑line tool is now live. The update upgrades the OpenAI Python library to the three point x series and swaps the HTTP client dependency from httpx to httpx two, addressing the installation breakage that prompted a quick zero point three two point one fix on Friday. New options let embedding functions accept a per‑call key, and the prompt command can now repeat templates so users can chain model configurations, such as pairing a high‑reasoning template with a custom SVG request. The tool also adds a reasoning summary switch for API models, offering concise or detailed summaries of model reasoning. This lets developers combine templates and pass keys to embedding plugins without changing shared state, widening flexibility for terminal‑based LLM workflows.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton