Nejumi 4 - M-IFEval (Japanese): leaderboard
Japanese prompts from M-IFEval (multilingual IFEval), whose instructions carry automatically verifiable format and content constraints, as run in Nejumi Leaderboard 4 (Weights & Biases Japan). Score is Nejumi's m_ifeval_score on a 0-100 scale; higher is better.
Metric: Score (%). Source: nejumi.ai. Saturation forecast: Around February 2027. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.5 (High) | 94.25 | #54 (Grok 4.5) |
| 2 | GPT-5.6 Sol (Max) | 92.48 | #22 (GPT-5.6 Sol) |
| 3 | Inkling (Max) | 92.04 | #144 (Inkling) |
| 4 | GPT-5.6 Terra (Max) | 91.59 | #40 (GPT-5.6 Terra) |
| 5 | Qwen 3.6 Max (Preview) (Thinking) | 91.59 | |
| 6 | DeepSeek V4.1 Flash | 91.15 | #37 |
| 7 | Muse Glimmer 30B (xHigh) | 90.71 | #191 (Muse Glimmer 30B) |
| 8 | GPT-5.4 (2026-03-05) (xHigh) | 90.71 | #63 (GPT-5.4 (2026-03-05)) |
| 9 | GPT-5.4 (2026-03-05) (High) | 90.71 | #63 (GPT-5.4 (2026-03-05)) |
| 10 | Gemini 3.1 Pro (Preview) | 90.27 | #64 |
| 11 | Gemini 3 Flash (Preview) | 90.27 | #90 |
| 12 | DeepSeek V4 Pro (Max) | 90.27 | #112 (DeepSeek V4 Pro) |
| 13 | Gemini 3.6 Flash | 89.38 | #57 |
| 14 | Muse Spark Open (xHigh) | 89.38 | #42 (Muse Spark Open) |
| 15 | GPT-5.2 (2025-12-11) (xHigh) | 89.38 |
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-m-ifeval-japanese · How It Works · Data refreshed daily, snapshot 2026-11-05.