Simon Willison has updated his LLM command-line tool and Python library to version 0.37, incorporating features developed over the last two months. The release introduces support for the new GPT-6.1 models, including gpt-6.1-sol and gpt-6.1-terra, alongside a command to simplify configuration of OpenAI-compatible endpoints. The default model has switched to gpt-6-luna, which costs half the price of the previous standard, gpt-5.6-luna. Embedding support now accepts varying content types, allowing integration with Gemini Embedding 2 for text, audio, images, and video.
Willison notes that twelve authors contributed fixes to this update, most of which were assisted by AI. He reports that reviewing these submissions was less burdensome than expected because he uses Opus 5.5 or GPT-6 Astra to triage the work. The contributors are listed as 0d00ciallo0721, ac1982, Allen Flux, Bian Yuhan, Bishop Kammeraad, Chen Yufeiyang, Christopher Krah, Ethan Stoner, Jiachen Xie, Olamide Faniyan, Sharat Chandra Sai Boddu, and nosmile99.
* Default model changed to gpt-6-luna for lower cost
* Gemini Embedding 2 support added for multimodal data
* New command facilitates OpenAI-compatible model setup
