ARC-AGI-2

A bird's-eye view of the benchmark: what it measures and every AI IQ chart built on it.

About ARC-AGI-2

ARC-AGI-2 measures abstract reasoning: each task is a small grid puzzle that shows a few input-output examples of a hidden transformation rule, and the model must infer the rule and apply it to a new input. The puzzles are designed to be novel, so scores reflect on-the-spot reasoning rather than memorized knowledge. Every result is scored per task with a published cost per attempt, taken from the official ARC Prize leaderboard.

ARC-AGI-2 Cost Efficiency
X = ARC Prize reported cost/task (log). Y = ARC-AGI-2 %. Each line connects one model's published reasoning-effort levels, so the score-vs-cost tradeoff is visible per model. Shows each model line's current and previous model. Color = provider.
Controls:

How to read this chart

Each line is one model run at several reasoning-effort levels on ARC-AGI-2. Every point shows the score and cost per task at that effort, with cheaper runs on the left. Up and to the left is better; a line that stays flat while moving right is buying little extra score with the extra spend. To keep the chart readable, each model line (e.g. Opus, Sonnet, Flash) shows only its current and previous model.

Data sources
ARC-AGI-2 Scores
Each model's ARC-AGI-2 score. Color = provider.
Controls:

How to read this chart

Bars rank models by the source-backed benchmark value used for this chart. Longer bars indicate higher published scores.

Data sources
ARC-AGI-2 vs Cost/Task
X = ARC Prize reported cost/task (log). Y = ARC-AGI-2 %. Color = provider.
Controls:
ARC-AGI-21:1Cost

How to read this chart

Each point is a public model. The chart compares ARC-AGI-2 % against Cost/Task, with color showing the model provider.

Data sources