Hacker News Reader

I wish they posted detailed metrics and benchmarks with such a "big" (loud) update.

The current livestream listed the benchmarks (curiously comparing it only to previous GPT models and not competitors)