scaling01 admits his Google prediction was wrong as Gemini catches the frontier again
scaling01 · x · 2026-11-21
Responding to a question about whether he has revised his Google estimate, prominent AI observer scaling01 admits his prediction was wrong.
- He still doesn't think Google's models beat the best OpenAI or Anthropic models, but concedes Google has "caught up again."
- He attributes the miss to expecting a much stronger runaway effect at the frontier; in reality it seems you can "build loopers and fix the RL slop" and get back to the frontier.
Takeaway: the frontier race is more contestable than assumed — leads don't snowball into an insurmountable moat.
Related event: scaling01 admits error: Google has caught up with OpenAI and Anthropic(3 posts)→
More from Models
- Tencent Hy4 ships first system card from a Chinese frontier lab, with third-party eval — S_OhEigeartaigh · 2026-11-21
- Tencent Hy4 Ships First Standalone Safety System Card With Third-Party Evals — wfithian · 2026-11-21
- Kimi K3.1 Pro jumps from #37 to #9 on Arena Alignment Index, deceptive completion down to 8.6% — arena · 2026-11-21
- Moonshot ties Anthropic for #2 lab on Arena Alignment Index, tops Chinese labs — arena · 2026-11-21
- Arena Alignment Index update: OpenAI sweeps top seven, Claude Fable 5.2 debuts at #8 — arena · 2026-11-21
- Asking Astra 6.2 to Review Code Written by Opus 5.5 Yields Almost Nothing — rickasaurus · 2026-11-21