Nejumi 4 - ARC-AGI-1: leaderboard
Metric: Accuracy (%). Source: nejumi.ai. 137 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (xHigh) | 98 |
| 2 | Gemini 3.6 Flash | 98 |
| 3 | GPT-5.6 Sol (Max) | 98 |
| 4 | GPT-5.4 (xHigh) | 98 |
| 5 | Kimi K3 (Max) | 98 |
| 6 | GPT-5.6 Luna (Max) | 98 |
| 7 | GPT-5.4 (High) | 98 |
| 8 | GPT-5.2 (xHigh) | 98 |
| 9 | Grok 4.20 0309 (Reasoning) | 98 |
| 10 | Claude Opus 5 (Max) | 98 |
| 11 | DeepSeek V4 Pro (Max) | 98 |
| 12 | Claude Opus 4.6 (Thinking) | 98 |
| 13 | DeepSeek V4.1 Flash | 98 |
| 14 | Claude Opus 4.7 (xHigh) | 98 |
| 15 | DeepSeek V4 Flash (0731) (Max) | 98 |
Interactive version: theaggregate.ai/benchmark?slug=nejumi-4-arc-agi-1 · How It Works · Data refreshed daily, snapshot 2026-09-19.