Internal benchmark chart fuels speculation Anthropic and OpenAI are a generation ahead internally
haider1 · x · 2026-11-09
Reading METR's AECI chart that includes internal Anthropic models, haider1 speculates: Anthropic likely used internal Model 3 to get Opus 5.5 to 'Astra-level' performance — something Model 2 couldn't do with fable 5.1 — and OpenAI is similarly holding back a Model 3 equivalent, suggesting both may be a generation ahead internally. The chart shows benchmark scores accelerating at 1.9× the pre-Oct 2025 rate (+29/yr vs +15/yr, 95% CI 1.5×–2.8×) once internal models are included, with Model 4 at 178.6 and Opus 5.5 at 174.6. Unconfirmed speculation.
More from Models
- KOL: Sora remains the only video model that generates indistinguishably real video — Kyrannio · 2026-11-09
- Hands-on Hermes Agent review: model-neutral, runs GLM 5.3 and beats Codex for GPT — Scobleizer · 2026-11-09
- Sora 2's Early Outputs Now Inaccessible at Any Price, Devs Lament Lost 'Taste' — gandamu_ml · 2026-11-09
- Users report GPT-6.1 Sol secretly routing to unreleased GPT-6.2 model — koltregaskes · 2026-11-09
- Thinking Machines Launches Inkling 2 Small: 594B MoE, Apache 2.0 Weights Next Week — geoffwolfe · 2026-11-09
- Claude Opus 5.5 Picked a Completely Random Test Secret on Its Own — davidcrawshaw · 2026-11-09