Summary & Key Takeaways ​

• llm-gemini 0.34 now supports the newly released Gemini 3.8 Flash model.
• Simon Willison highlights Gemini 3.8 Flash's speed, low cost, and proficiency in HTML/JavaScript.
• He demonstrates the model generating a "cool thing in HTML" in 13 seconds for 1.8 cents.
• The model was used to add sandboxed HTML rendering to his markdown-svg-renderer tool.
• The post includes a transcript of the interaction with the AI for this development.

Our Commentary ​

Simon Willison always delivers the goods with his practical LLM explorations. Seeing Gemini 3.8 Flash praised for its HTML/JS competence, speed, and low cost is genuinely exciting. I love that he immediately put it to work improving his markdown-svg-renderer. This is the kind of real-world validation that makes me want to dive in and try it myself.