Independent analysis cross-checks METR report: Anthropic internal models show ~1.9x capability acceleration
RishiBommasani · x · 2026-11-08
Karthik Tadepalli cross-checked the non-redacted parts of the METR report against public data, finding the public data line up with METR's estimates of capability acceleration on Anthropic's internal models. The chart shows four estimates: public ECI (Bayesian) 1.42x speedup (95% CI 1.24–1.66), public ECI (METR frequentist) 1.64x (1.37–1.99), Anthropic ECI on public models 1.76x (1.34–2.35), and Anthropic ECI on internal models 1.91x (1.50–2.80). The takeaway: Anthropic's AI acceleration is probably real, with the open question being whether bottlenecks will slow the trend.
More from Research
- The Formula Behind ChatGPT Was Invented in 1847 to Calculate Planetary Orbits — burny_tech · 2026-11-08
- NeurIPS 2026 October newsletter: Sydney Dec 6-12, grants and invited talks announced — NeurIPSConf · 2026-11-08
- Tsinghua paper: hidden-goal agents evade top CoT monitors ~60% of the time — rohanpaul_ai · 2026-11-08
- Subbarao Kambhampati talks 'The Mythos of LLM Thinking Traces' at University of Bordeaux — rao2z · 2026-11-08
- DeepMind's MuJoCo Physics Simulator Passes 16k GitHub Stars, Widely Used in Robotics — google-deepmind · 2026-11-08
- Looped Transformers explained: when is another pass worth the cost? — TheTuringPost · 2026-11-08