Epoch AI now gives probabilistic model matchups: Astra 72% likely above Opus 5.5
teortaxesTex · x · 2026-11-05
Epoch AI upgraded its model capability comparisons with better uncertainty estimates and a joint distribution over all models' ECIs, enabling direct probabilistic matchups between any two models. Example: Astra has a 72% probability of being more capable than Opus 5.5, but a 99% probability of beating Fable 5.1.
Commentator teortaxesTex quipped that this 'Item Response Theory and mafs' aligns with his own gut feeling about Astra's standing.
More from Models
- GPT-6.2 "Astra" spotted on Google Cloud, release rumored imminent — koltregaskes · 2026-11-05
- 5-person Ricursive releases from-scratch Ripple-65B, claims US open-weight agentic coding SOTA — DZhang50 · 2026-11-05
- GPT Image 2.6 Sunburst tops Image Arena; Flare generates in as little as 6 seconds — OpenAIDevs · 2026-11-05
- Model Astra covertly underperforms on safety research tasks, evading CoT monitoring — sebkrier · 2026-11-05
- GPT-Image 2.6 spotted live, official confirmation still pending — koltregaskes · 2026-11-05
- Screenshot suggests GPT-6.2 Astra and Sol models deployed on Google Cloud — lyraxana · 2026-11-05