Nejumi 4 - MT-Bench (Japanese) - Writing: leaderboard
Japanese MT-Bench questions in the writing category, two-turn answers rated 1-10 by GPT-4.1 as judge in Nejumi Leaderboard 4 (Weights & Biases Japan). Nejumi stores ratings divided by 10; the board multiplies by 100 (10-100), the scale of Nejumi 4 - MT-Bench (Japanese). Higher is better.
Metric: Judge rating (1-10, x10). Source: nejumi.ai. Saturation forecast: Estimated already saturated. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Flash Lite (Preview) | 99.5 | #201 |
| 2 | mistral-large-4-oss (High) | 99.5 | #60 (mistral-large-4-oss) |
| 3 | Claude Sonnet 4.6 (Thinking) | 99.5 | #94 (Claude Sonnet 4.6) |
| 4 | Muse Spark Open (xHigh) | 99.5 | #42 (Muse Spark Open) |
| 5 | Kimi K3 (Thinking) | 99.5 | #35 (Kimi K3) |
| 6 | Qwen 3.6 Max (Preview) (Thinking) | 99 | |
| 7 | Inkling (Max) | 99 | #144 (Inkling) |
| 8 | DeepSeek V3.2 (Thinking) | 98.5 | #210 (DeepSeek V3.2) |
| 9 | Claude Opus 4.6 (Thinking) | 98.5 | #67 (Claude Opus 4.6) |
| 10 | GPT-5.6 Luna (Max) | 98.5 | #79 (GPT-5.6 Luna) |
| 11 | GPT-5 (2025-08-07) (High) | 98.5 | |
| 12 | Claude Opus 4.7 (xHigh) | 98.5 | #51 (Claude Opus 4.7) |
| 13 | GPT-5.4 (2026-03-05) (xHigh) | 98.5 | #63 (GPT-5.4 (2026-03-05)) |
| 14 | Qwen 3.6 35B A3B (Thinking) | 98.5 | #238 (Qwen 3.6 35B A3B) |
| 15 | GPT-5.1 (2025-11-13) (High) | 98.5 | #107 (GPT-5.1 (2025-11-13)) |
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-mt-bench-japanese-writing · How It Works · Data refreshed daily, snapshot 2026-11-05.