Benchmark Radar™
Star —

ARC-AGI Benchmark

Explore score history ↗

The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI) is a benchmark designed to test general intelligence and abstract reasoning capabilities through visual grid-based transformation tasks. Each task consists of 2-5 demonstration pairs showing input grids transformed into output grids according to underlying rules, with test-takers required to infer these rules and apply them to novel test inputs. The benchmark uses colored grids (up to 30x30) with 10 discrete colors/symbols, designed to measure human-like general fluid intelligence and skill-acquisition efficiency with minimal prior knowledge.

Released
2019-11-05
Modality
image
Evaluates
reasoning, spatial reasoning, vision
Openness
unknown
Importer
LLM Stats

Source provenance

ARC-AGI results and reported scores

Reported benchmark scores · LLM Stats

Reported scores: 8

ModelOrganizationReported valueReportedEvidence
GPT-5.2OpenAI0.8622025-12-11evidence
GPT-5.2 ProOpenAI0.9052025-12-11evidence
GPT-5.4OpenAI0.9372026-03-05evidence
GPT-5.5OpenAI0.952026-04-23evidence
Inkling-SmallThinking Machines Lab0.842026-07-30evidence
LongCat-Flash-ThinkingMeituan0.5032025-09-22evidence
o3OpenAI0.882025-04-16evidence
Qwen3-235B-A22B-Instruct-2507Qwen0.4182025-07-22evidence

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?

Related reasoning, spatial reasoning, vision benchmarks