Nejumi 4 - jaster (2-shot) - JSQuAD: leaderboard
JSQuAD (JGLUE), Japanese extractive reading comprehension, from the llm-jp-eval jaster suite as run 2-shot by Nejumi Leaderboard 4 (Weights & Biases Japan). Score is character-level F1 x100 on Nejumi's sampled test split; higher is better.
Metric: Character F1 (x100). Source: nejumi.ai. Saturation forecast: Estimated already saturated. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 Next 80B A3B Instruct | 82.64 | #381 |
| 2 | Inkling (Max) | 82.28 | #144 (Inkling) |
| 3 | Claude Opus 5 (Max) | 82.11 | #16 (Claude Opus 5) |
| 4 | GLM-5.3 Flash (Max) | 82.06 | #77 (GLM-5.3 Flash) |
| 5 | Qwen 3.6 27B (Thinking) | 81.92 | #182 (Qwen 3.6 27B) |
| 6 | Gemini 3.6 Flash | 81.91 | #57 |
| 7 | Gemini 3.1 Pro (Preview) | 81.81 | #64 |
| 8 | MiMo-V2.5-Pro (Thinking) | 81.56 | #167 (MiMo-V2.5-Pro) |
| 9 | Claude Opus 4 (Non-reasoning) | 81.5 | #170 (Claude Opus 4) |
| 10 | Grok 4.5 (High) | 81.42 | #54 (Grok 4.5) |
| 11 | Gemini 3.5 Flash | 81.39 | #66 |
| 12 | Gemini 3 Pro (Preview) | 81.31 | #71 |
| 13 | Hy4 preview (High) | 81.3 | #45 (Hy4 preview) |
| 14 | GPT-5.4 Mini (2026-03-17) (High) | 81.05 | #180 (GPT-5.4 Mini (2026-03-17)) |
| 15 | Claude Opus 4.5 (Non-reasoning) | 81.03 | #87 (Claude Opus 4.5) |
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-jsquad · How It Works · Data refreshed daily, snapshot 2026-11-05.