Nejumi 4 - GLP - Reasoning: leaderboard
An aggregate of Nejumi 4's three reasoning sub-scores: abstract reasoning on ARC-AGI grid puzzles, logical reasoning on the ten MT-Bench reasoning questions, and mathematical reasoning on MAWPS, MGSM and the MT-Bench math questions. The board is the unweighted mean of the three in percent, higher is better. Source: github.com/wandb/llm-leaderboard.
Metric: Score (%). Source: nejumi.ai. Saturation forecast: Estimated already saturated. 139 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 | 96.56 | #12 |
| 2 | Claude Fable 5 | 96.33 | #16 |
| 3 | GPT-5.6 Terra | 94.33 | #41 |
| 4 | Qwen3.8 2.4T A95B | 94.11 | #55 |
| 5 | GPT-5.6 Sol | 92.78 | #19 |
| 6 | Claude Opus 4.8 | 91.67 | #37 |
| 7 | Gemini 3.6 Flash | 91.61 | #51 |
| 8 | Gemini 3.1 Pro (Preview) | 91.28 | #58 |
| 9 | Claude Opus 4.7 | 90.94 | #49 |
| 10 | Claude Opus 4.6 | 90.78 | #63 |
| 11 | DeepSeek V4.1 Flash | 90.61 | #36 |
| 12 | GPT-5.5 (2026-04-23) | 90.44 | #22 |
| 13 | GPT-5.4 (2026-03-05) | 89.94 | #59 |
| 14 | Hy4 preview | 89.28 | #38 |
| 15 | Inkling Small | 89.22 | #138 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=nejumi-4-glp-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.