Simon Willison's blog offers a mix of technical tutorials, data analysis projects, and reflections on technology and society. With a focus on Python, SQL, and web development, Simon shares insights into building web applications, working with data, and exploring emerging technologies. Developers can learn about data visualization, API design, and software engineering best practices, gaining inspiration and practical knowledge to advance their careers.

Simon Willison

Simon Willison describes porting the Moebius 0.2B image inpainting model to run entirely in the browser using ONNX Runtime Web and WebGPU, with Claude Code doing virtually all the coding work. Starting from a PyTorch/CUDA-only model, Claude Code converted the weights to ONNX format, published them to Hugging Face, and built a web UI hosted on GitHub Pages. Key findings: Claude Opus 4.8 can autonomously handle PyTorch-to-ONNX conversion and full web app deployment; Chrome, Firefox, and Safari all support running this model; and the CacheStorage API can handle ~1.3GB model files for browser-side caching. The author used a parallel-agent workflow, running Claude Code on this side project while Codex Desktop worked on a separate main project.

Porting the Moebius 0.2B image inpainting model to run in the browser with Claude Code