Claude Opus 5 reaches 30.2% on ARC-AGI-3, far above prior frontier scores
EricBuess · x · 2026-07-25
A repost says ARC-AGI-3 was designed to be brutally hard: humans solve all environments, while as of March every frontier model was below 1%.
The attached chart shows Claude Opus 5 (high) reaching 30.2%, far above the previous reported frontier result of 7.8%. It also visualizes the cost-to-score tradeoff, with Opus 5 sitting near the high-cost, high-score end of the curve.
Related event: Claude Opus 5 Sets New ARC-AGI-3 Record(15 posts)→
More from Models
- DeepSeek cut KV cache per token from 389KB to 890B in two years as open-closed gap debate rages — teortaxesTex · 2026-10-27
- Unverified: fable 5.5 speculated at 4-5T parameters to outscale 1-3T Opus — haider1 · 2026-10-27
- Kimi K3 Is 'Open' Until You Succeed: License Caps Free Use at $20M Revenue — alex_verem · 2026-10-27
- Screenshot shows 'GPT-6 Astra Ultra' in ChatGPT model picker — unverified leak or joke — justalexoki · 2026-10-27
- DeepSeek Safety Eval Finds "Over-Attribution": Model Mistakes Ordinary Tasks for Tests — Much_Boysenberry9124 · 2026-10-27
- Custom VQA benchmark under 48GB VRAM: Qwen3.8-Flash-Next leads, Qwen3.8-27B regressed — EmilPi · 2026-10-27