Anthropic's internal AECI index suggests minimal gains for next model
ChrisGPT · x · 2026-08-21
Anthropic inadvertently revealed the performance of its upcoming model on its internal Anthropic ECI (AECI) index. AECI is Anthropic's version of Epoch's Capabilities Index, compressing various benchmark results into a single score using internal data. Mythos 5 lands around 161 AECI, while Model 2 is only 1.5 points higher (162.5 - 163), with large error bars. However, a larger gap on CoBench suggests Anthropic is currently focusing on scaling internal researcher capabilities rather than broader AECI metrics.
More from Models
- Developer Warns Uncensored Qwen 3.8 27B Model on Mac Immediately Explains How to Make Meth — Polymarket · 2026-08-21
- Adding Mermaid Support Becomes a Touchstone for Model Capabilities — oran_ge · 2026-08-21
- Liquid AI releases DSpark draft models, boosting inference speed by up to 3.18x — JosephJacks_ · 2026-08-21
- Opinion: Fable 5 is the only usable model in Anthropic's latest generation — bindureddy · 2026-08-21
- Autoregressive models can beat diffusion in image generation — cloneofsimo · 2026-08-21
- Critique of GeneralistAI Demo: Lack of OOD Evidence in Behavior Prompting — chris_j_paxton · 2026-08-21