Nejumi 4 - jaster (2-shot) - JMMLU: leaderboard
JMMLU, Japanese multiple-choice exam questions across academic subjects, from the llm-jp-eval jaster suite as run 2-shot by Nejumi Leaderboard 4 (Weights & Biases Japan). Score is exact match (%) on Nejumi's sampled test split; higher is better.
Metric: Exact match (%). Source: nejumi.ai. Saturation forecast: Estimated already saturated. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 97 | #64 |
| 2 | Claude Opus 4.8 (xHigh) | 97 | #39 (Claude Opus 4.8) |
| 3 | Gemini 3.5 Flash | 96 | #66 |
| 4 | Claude Sonnet 4.5 (Thinking) | 96 | #148 (Claude Sonnet 4.5) |
| 5 | GPT-5.6 Sol (Max) | 96 | #22 (GPT-5.6 Sol) |
| 6 | Claude Opus 5 (Max) | 96 | #16 (Claude Opus 5) |
| 7 | GPT-5.6 Terra (Max) | 96 | #40 (GPT-5.6 Terra) |
| 8 | Grok 4.5 (High) | 96 | #54 (Grok 4.5) |
| 9 | Gemini 3 Flash (Preview) | 95 | #90 |
| 10 | Gemini 3.6 Flash | 95 | #57 |
| 11 | Kimi K3 (Max) | 95 | #35 (Kimi K3) |
| 12 | GPT-5.6 Luna (Max) | 95 | #79 (GPT-5.6 Luna) |
| 13 | Qwen 3.6 Max (Preview) (Thinking) | 95 | |
| 14 | GPT-5.4 (2026-03-05) (High) | 95 | #63 (GPT-5.4 (2026-03-05)) |
| 15 | Kimi K3 (Thinking) | 95 | #35 (Kimi K3) |
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-jaster-2-shot-jmmlu · How It Works · Data refreshed daily, snapshot 2026-11-05.