SimonWillison.net is the long-running blog of Simon Willison, a software engineer, open-source creator, and co-author of the original Django framework. He writes about Python, Django, Datasette, AI tooling, prompt engineering, search, databases, APIs, data journalism, and practical software architecture. The blog includes detailed notes from experiments, conference talks, and real projects. Readers will find clear explanations of topics such as LLM workflows, SQL patterns, data publishing, scraping, deployment, caching, and modern developer tooling. Simon also publishes frequent micro-posts and TIL entries that document small discoveries and tricks from day-to-day engineering work. The tone is practical and research oriented, making the site a valuable resource for anyone interested in serious engineering and open data.
213 articles from this blog
A guide to parakeet-mlx, a project porting NVIDIA's Parakeet ASR model to Apple's MLX framework for fast, local audio transcription.
Analysis of GPT-5.1's new adaptive thinking features, model routing system, and safety benchmarks from the system card addendum.
OpenAI releases GPT-5.1 API with new reasoning modes, adaptive reasoning, extended prompt caching, and new built-in tools for developers.
Datasette 1.0a22 release notes covering new security features, a client detection method, and developer tools for plugin authors.
A deep dive into Google's Nano Banana (Gemini 2.5 Flash) AI image model, exploring its autoregressive architecture and superior prompt engineering capabilities.
OpenAI objects to court order demanding 20M ChatGPT user conversations, citing dangerous precedent for AI discovery.
A humorous look at AI model benchmarking using the challenge of generating an SVG of a pelican riding a bicycle, and the risks of labs 'gaming' the test.
Explains how MCP servers enable faster development by using LLMs to dynamically read specs, unlike traditional APIs.
A clever hack using POSIX advisory locks for cross-container communication between processes on the same machine.
An analysis of scaling HNSW vector indexing in Redis, covering new contributions for efficient deletions and parallel queries across distributed nodes.
Experiment testing if AI vision models improve SVG drawings of a pelican on a bicycle through iterative, agentic feedback loops.
Using AI coding agents to automate repetitive plugin upgrades for Datasette 1.0, running six parallel sessions.
Netflix's guidelines for using generative AI in content production, focusing on copyright, data security, and talent rights.
Testing various LLMs to generate a POV-Ray script for a pelican riding a bicycle, comparing results and fixing errors.
A developer reverse-engineers the Codex CLI to directly prompt the new GPT-5-Codex-Mini model, detailing the process in a technical blog post.
Analysis of MCP's advantages over OpenAPI, focusing on secure authentication and OAuth dynamic client registration for AI agents.
Mastodon 4.5 introduces official quote post support and automatically fetches all replies to fix conversation gaps in the Fediverse.
A developer with AiDHD discusses the challenges of focus and feature creep when building MVPs with AI, emphasizing the need for discipline.
Explores how LLMs could lower the barrier to creating and adopting new programming languages by handling syntax and core concepts.
Guide to running OpenAI's Codex CLI with a self-hosted LLM on an NVIDIA DGX Spark via Tailscale for remote coding assistance.
