ARC-AGI Benchmark

Explore score history ↗

ARC-AGI is an AI benchmark in the Benchmark Radar catalog.

Released: 2019-11-06
Openness: unknown
Importer: Benchmark Saturation

Source provenance

Original evidence

ARC-AGI results and reported scores

Reported benchmark scores · Benchmark Saturation

Reported scores: 5

Model	Organization	Reported value	Reported	Evidence
Claude Sonnet 4.5	—	63.7	—	evidence
E. Pang	—	77.1	—	evidence
GPT-5 Pro	—	70.2	—	evidence
Grok 4	—	66.7	—	evidence
J. Berman	—	79.6	—	evidence

Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.

Which sources cite ARC-AGI?

ARC-AGI