llm-gemini 0.34
Simon Willison announced the release of llm-gemini 0.34, an LLM plugin that provides access to Google's Gemini family of models. The key addition is support for the new gemini-3.8-flash model, corresponding to Google's Gemini 3.8 Flash release, which offers configurable thinking levels: low, medium, and high. The release also includes a fix for async responses failing to record the resolved model version, with thanks to contributor Charlie Tonneslan. Willison highlights that Gemini 3.8 Flash is fast, cheap, and competent at tasks like generating HTML and JavaScript. He demonstrates this by prompting it to create a 'cool thing in html,' which took 13 seconds and cost 1.8 cents. Additionally, he used Gemini 3.8 Flash with his llm-coding-agent plugin to extend his markdown-svg-renderer tool to render HTML blocks in Markdown using a sandboxed iframe. The source also notes that Google released Gemini 3.8 Flash Cyber, but that version is available only to 'trusted defenders.'
Developers using llm-gemini can now access Gemini 3.8 Flash, which offers fast, low-cost generation for HTML and JavaScript tasks.