LLM Reference
ARC-AGI-2 HighactiveReasoningVision

ARC-AGI-2 High: ARC-AGI-2 — High Effort

Metric: Accuracy (higher is better)Introduced: 2026

High-effort ARC-AGI-2 variant for abstract visual pattern reasoning. Kept separate from general ARC-AGI-2 rows that may use different effort settings.

Models ranked

2

tracked on this benchmark

Score band

83.3 – 72.1

best → lowest tracked

Snapshot trend

-11.20

Apr 24 → May 28 · 1 models

Leaderboard

Tracked models ranked by Accuracy (higher is better).

Compare candidates
#Model variant and provenanceScore
1
GPT-5.5 Pro
Version: ARC-AGI-2 high effortHarness: Not recordedEvaluator: Not recordedObserved: Apr 24, 2026Confidence: Not recordedSource

Notes: High-effort ARC-AGI-2 variant; kept separate from generic ARC-AGI-2 rows.

83.3
2
Claude Opus 4.8
Version: ARC-AGI-2 high effortHarness: Not recordedEvaluator: Not recordedObserved: May 28, 2026Confidence: Not recordedSource

Notes: High-effort ARC-AGI-2 variant from secondary source.

72.1

How to read this benchmark

This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.

Trust this score when

  • There is a fresh timestamped snapshot (or multiple snapshots) for this benchmark.
  • The model list covers the same version family you can actually deploy today.
  • Top candidates overlap with your required routing and feature requirements.

Be cautious when

  • There is only one benchmark snapshot or the dataset appears stale.
  • The benchmark metric direction is opposite of your decision objective.
  • The score difference between options is narrow and likely within implementation variance.

Related benchmarks

GPQA
Google-Proof Q&A
DROP
Discrete Reasoning Over Paragraphs
ARC
AI2 Reasoning Challenge
HellaSwag
MMMU Pro
ANLS*
Average Normalized Levenshtein Similarity*

Last reviewed: Jun 17, 2026

Resources

Website