Hacker News Reader

It seems weird that a likely prime indicator of capability isn't mentioned, the model size.

I thought most frontier LLM providers don’t disclose exact parameter counts these days.

No, but open models do along with their architectures, and other technical training details.