Nejumi 4 - jaster (2-shot) - JSeM: leaderboard
JSeM, the Japanese semantic test suite of entailment problems, 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: Around April 2027. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 84 | #64 |
| 2 | Gemini 3.5 Flash | 84 | #66 |
| 3 | GLM-5.3 (Max) | 84 | #44 (GLM-5.3) |
| 4 | Claude Opus 5 (Max) | 84 | #16 (Claude Opus 5) |
| 5 | Kimi K3 (Max) | 83 | #35 (Kimi K3) |
| 6 | Qwen 3 Max (Preview) | 83 | #248 |
| 7 | Qwen 3.6 27B (Thinking) | 83 | #182 (Qwen 3.6 27B) |
| 8 | Gemini 3 Flash (Preview) | 82 | #90 |
| 9 | Gemma 4 26B A4B (IT) | 82 | #293 |
| 10 | O4 Mini (2025-04-16) | 82 | #183 |
| 11 | GPT-5.6 Sol (Max) | 82 | #22 (GPT-5.6 Sol) |
| 12 | Claude Opus 4.5 (Thinking) | 82 | #87 (Claude Opus 4.5) |
| 13 | Qwen 3.6 Max (Preview) (Thinking) | 82 | |
| 14 | GPT-5.4 (2026-03-05) (High) | 82 | #63 (GPT-5.4 (2026-03-05)) |
| 15 | MiniMax-M3 (Thinking) | 82 | #151 (MiniMax-M3) |
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-jsem · How It Works · Data refreshed daily, snapshot 2026-11-05.