Nejumi 4 - ARC-AGI-2: leaderboard
Metric: Accuracy (%). Source: nejumi.ai. 137 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Terra (Max) | 100 |
| 2 | GPT-5.6 Sol (Max) | 96 |
| 3 | Claude Opus 5 (Max) | 94 |
| 4 | GPT-5.5 (xHigh) | 88 |
| 5 | GPT-5.4 (xHigh) | 84 |
| 6 | Claude Opus 4.8 (xHigh) | 78 |
| 7 | Gemini 3.1 Pro (Preview) | 72 |
| 8 | Claude Opus 4.6 (Thinking) | 72 |
| 9 | Gemini 3.6 Flash | 70 |
| 10 | GPT-5.4 (High) | 68 |
| 11 | Claude Opus 4.7 (xHigh) | 68 |
| 12 | Grok 4.20 0309 (Reasoning) | 64 |
| 13 | DeepSeek V4.1 Flash | 64 |
| 14 | DeepSeek V4 Flash (0731) (Max) | 62 |
| 15 | Gemini 3.5 Flash | 60 |
Interactive version: theaggregate.ai/benchmark?slug=nejumi-4-arc-agi-2 · How It Works · Data refreshed daily, snapshot 2026-09-19.