Nejumi 4 - GLP Average: leaderboard
The capability half of Nejumi 4, the unweighted mean of its five second-level categories: applied language, reasoning, knowledge and question answering, foundational language, and application development. Those cover expression, translation, reading comprehension, grid puzzles, maths, knowledge exams, inference, grammaticality, coding and function calling. Higher is better. Source: github.com/wandb/llm-leaderboard.
Metric: Score (%). Source: nejumi.ai. Saturation forecast: Around January 2027. 139 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 | 86.68 | #12 |
| 2 | Claude Fable 5 | 84.99 | #16 |
| 3 | GPT-5.6 Sol | 84.02 | #19 |
| 4 | GPT-5.6 Terra | 83.63 | #41 |
| 5 | Claude Opus 4.8 | 83.28 | #37 |
| 6 | GPT-5.5 (2026-04-23) | 83.03 | #22 |
| 7 | Claude Opus 4.7 | 82.99 | #49 |
| 8 | Qwen3.8 2.4T A95B | 82.9 | #55 |
| 9 | Gemini 3.1 Pro (Preview) | 82.81 | #58 |
| 10 | Gemini 3.6 Flash | 82.47 | #51 |
| 11 | Kimi K3 | 82 | #30 |
| 12 | Gemini 3.5 Flash | 81.53 | #60 |
| 13 | Hy4 preview | 81.43 | #38 |
| 14 | DeepSeek V4.1 Flash | 81.35 | #36 |
| 15 | Claude Opus 4.6 | 81.25 | #63 |
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-average · How It Works · Data refreshed daily, snapshot 2026-10-20.