Greg Kamradt evaluated the reasoning logs using AmpCode agents.
François Chollet#37
Anthropic Opus 4.8 is new SOTA on ARC-AGI-3
Score: 1.5%, ~$10K
ARC-AGI-3 analysis notes: * Opus 4.8 read the environment an abstraction *above* Opus 4.7, as objects & systems, not pictures * Opus 4.8 succeeded on early levels, but still committed to a wrong sub-goal
Opus 4.8 just broke ARC-AGI-3
it tripled GPT-5.5's score
we are now at a breathtaking 1.5% human efficiency
Greg Kamradt evaluated the reasoning logs using AmpCode agents.
François Chollet#37Anthropic Opus 4.8 is new SOTA on ARC-AGI-3
Score: 1.5%, ~$10K
ARC-AGI-3 analysis notes: * Opus 4.8 read the environment an abstraction *above* Opus 4.7, as objects & systems, not pictures * Opus 4.8 succeeded on early levels, but still committed to a wrong sub-goal
Positive users praise Claude Opus 4.8's incremental ARC-AGI-3 gains and knowledge while negative users criticize exaggerated claims, high costs, and irrelevant benchmarks.
