ARC-AGI-3

Interactive benchmark of novel game-like environments that agents must explore and learn without instructions.

Reference by ARC Prize Foundation

IndexedLicense unknown

About

Whether an agent can explore an unfamiliar environment, infer its goals and rules, and keep learning as it plays, compared with humans.

Use cases: llm-evaluation

Reviews

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

About the benchmark

Benchmark site: arcprize.org/arc-agi/3
Maintainer: ARC Prize Foundationarcprize.org
Launched: 2026-03-25arcprize.org
Paper: arcprize.org/media/ARC_AGI_3_Technical_Report.pdf
Leaderboard: arcprize.org/leaderboard

We link to the official leaderboard instead of copying scores. Benchmarks get gamed and saturate; we never use them to rank listings.

Details

License: License unknown
Website: arcprize.org/
Docs: arcprize.org/arc-agi/3
First seen: Oct 24, 2026
Last checked: Oct 24, 2026
Source: Specification Website Paper Leaderboard

More from ARC Prize Foundation

0.501score
ARC-AGI-2
ARC Prize Foundation
Grid puzzles that are easy for people but hard for AI, testing abstraction and reasoning on new problems.
ReferenceFFreshness: F★ 746

Cite this page

ARC-AGI-3 benchmark. Agent Mart. https://agentmart.dev/reference/arc-agi-3 (last updated 2026-10-24)