Nejumi 4 - jaster (0-shot) - JBLiMP: leaderboard
JBLiMP, choosing the grammatical sentence of Japanese linguistic minimal pairs, from the llm-jp-eval jaster suite as run 0-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 | Claude Opus 5 (Max) | 98 | #16 (Claude Opus 5) |
| 2 | GPT-5.4 (2026-03-05) (xHigh) | 98 | #63 (GPT-5.4 (2026-03-05)) |
| 3 | Gemini 3 Pro (Preview) | 97 | #71 |
| 4 | Gemini 3.5 Flash | 97 | #66 |
| 5 | GPT-5.6 Sol (Max) | 97 | #22 (GPT-5.6 Sol) |
| 6 | Kimi K2.5 (Thinking) | 97 | #152 (Kimi K2.5) |
| 7 | DeepSeek V4 Flash (Max) | 97 | #155 (DeepSeek V4 Flash) |
| 8 | GLM-5.3 Flash (Max) | 97 | #77 (GLM-5.3 Flash) |
| 9 | GPT-5.4 (2026-03-05) (High) | 97 | #63 (GPT-5.4 (2026-03-05)) |
| 10 | Gemini 3.1 Pro (Preview) | 96 | #64 |
| 11 | DeepSeek V4.1 Flash | 96 | #37 |
| 12 | Gemini 3.6 Flash | 96 | #57 |
| 13 | DeepSeek V4 Pro (Max) | 96 | #112 (DeepSeek V4 Pro) |
| 14 | Grok 4.5 (High) | 96 | #54 (Grok 4.5) |
| 15 | Grok 4.20 0309 (Reasoning) | 96 | #120 (Grok 4.20 0309) |
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-0-shot-jblimp · How It Works · Data refreshed daily, snapshot 2026-11-05.