Hacker News Reader

They didn't show this, they just increased the length where accuracy breaks down.

Explain? OpenAI showed the new scaling law in December 2024 that performance keeps increasing proportional to ln(N reasoning tokens)

link?