ARC-AGI-2 A
AI benchmark · Reasoning
What is ARC-AGI-2?
ARC-AGI-2 measures fluid intelligence and abstract reasoning in AI systems through novel input-output grid puzzles calibrated against human performance.
General capability mostly
“specifically designed to provide a more granular signal to assess abstract reasoning and problem-solving abilities at higher levels of fluid intelligence”
arxiv.org/abs/2505.11831 ↗
Who built ARC-AGI-2
Led by ARC Prize Foundation.
Paper: arxiv.org/abs/2505.11831 ↗
Lineage
Benchmarks this one is built on and built from: versions, subsets, extensions and the indexes that combine them.
Built on
ARC-AGI-1 earlier version
Derived benchmarks
ARC-AGI-3 later version
Tier A
Widely used: a lab reports it, or 50 or more models are scored on it.
How much stands behind a tier: usage, who backs it, a confirmed line from its own site or paper, and a clear read on X. High with three of the four, medium with two, low otherwise.
- 109 models scored
- in curated RL directories (2 pts)
On X · mixed, moderate buzz, from 13 posts about ARC-AGI-2 in the last month
Still used for frontier model results and the ARC Prize 2026 competition, but scores near 84–95% make many see it as nearly saturated, with attention shifting to ARC-AGI-3.
How we got to tier A
- Adoption: 109 models scored. Tier A.
- On X: mixed, moderate buzz. “Still used for frontier model results and the ARC Prize 2026 competition, but scores near 84–95% make many see it as nearly saturated, with attention shifting to ARC-AGI-3.” No move.
- In curated RL directories: 2 points, a floor of C.
- Confirmed: a line from its own site or paper places it.
A clear view on X moves a tier one step up or down; most entries don't move. How tiers work.
Where ARC-AGI-2 sits
Its domain on the map, live. Hover a logo for its name; click to open it. Open it on the full map
Next: browse all benchmarks. Entry from the RL Research daily scrape of public sources, 2026-11-05 snapshot.