Inference
ARC Prize Foundation

ARC-AGI-3 (semi-private)

Abstract reasoning puzzles held out from training data.

ModelsModels with a result.15
Top resultBest result on this test.N/A
Top-3 spreadPoints from first to third.N/A
YearYear of release.N/A

Result and price

020406080100$1$3$10$30$100RESULTOUTPUT PRICE PER 1M TOKENS · LOG SCALE

Results

15 results
#ModelResultThe score from the source.
—GPT-6 Astra - Provider Adaptermax effort98.6%
—GPT-6 Astramax effort62.7%
—Claude Opus 5high effort30.2%
—GPT-5.6 Solmax effort7.8%
—Grok 4.6xhigh effort2.1%
—Claude Opus 4.8high effort1.5%
—GPT-5.6 Terramax effort0.8%
—Anthropic Opus 4.6max effort0.5%
—GPT-5.5high effort0.4%
—Gemini 3.1 Pro Preview0.4%
—Grok 4.5high effort0.3%
—GPT-5.4high effort0.2%
—GPT-5.6 Lunamax effort0.2%
—Opus 4.7high effort0.2%
—Grok 4.20 (Beta Reasoning)0.1%