Opus 5 hits 30% on ARC-AGI-3, but the gain does not transfer to Witness
Charuru · reddit · 2026-07-25
A screenshot of a Guanghan Ning post says Opus 5 scores 30% on ARC-AGI-3, roughly 4× the previous best and about 20× Opus 4.8. But on the held-out Witness suite, the gain does not transfer: Opus 5 ties kimi-k3 and Fable-5 overall, beats Opus 4.8, and still falls short of a true generational jump. The analysis argues the model looks strong on familiar puzzle templates but regresses on novelty and interactive rule discovery.
Related event: Claude Opus 5 Sets New ARC-AGI-3 Record(15 posts)→
More from Models
- OpenAI upgrades Decisions API: 400 questions per call, native voice and image support — OpenAIDevs · 2026-10-26
- Fangcun AI, China's first independent frontier AI safety lab, launches with DeepSeek eval report — JacquesThibs · 2026-10-26
- When AI hits an existential crisis: model confuses itself and everyone else — Ok-Delivery-7437 · 2026-10-26
- Kimi K3.1 Pro tops MathArena open models, closing frontier gap to about 6 months — teortaxesTex · 2026-10-26
- Ruff creator Charlie Marsh: frontier models find infinite bugs in any software — charliermarsh · 2026-10-26
- Dev testing one skill across Opus 5.5 and Sol 6.1 finds instruction following 'all over the map' — lucasmeijer · 2026-10-26