@yt's notes

GPT-5 came out underwhelming. Why this happened, and what awaits the industry now. Just trying to think this through.

Claude Code is the first AI coding tool that I feel is genuinely useful not just a cool tech demo, but something that I see myself using in my workflow daily. Here's how it helped my implement new features in a real-world repository.

In this article I throw my old demo tracks at Suno AI, to see if it can remake it in good quality with a different singer, and so on. The result is pretty impressive. Here's what I've found out.

GPT-4.1 released being a fast, but rather an underperforming model still behind Sonnet 3.7 and Gemini 2.5 Pro.

Anthropic has finally released a new Claude model, the one that can be used in both thinking and non-thinking mode. I, for one, plugged it into nicechat to test it out, and it's ok. My subjective feeling is that it's good, but o1 and o3 are still better when comes to following instructions.