What benchmarks miss about LLMs
Benchmarks do not tell the whole story.
Blog
Writing about code, tools, and the web.
Fast iteration beats big models
Kimi K2.5 Turbo and fast iteration. Why I prefer speed over waiting for big models.
What I like and worry about with AI
What works, what worries me, and the main rules I follow.
Creative Coding
Making pointless things with code is really cool.