Nejumi 4 - jaster (0-shot) - JCoLA (out-of-domain): leaderboard
JCoLA out-of-domain split, Japanese grammatical acceptability judgments, 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 | Gemini 2.5 Pro | 92 | #159 |
| 2 | Gemini 3 Pro (Preview) | 92 | #71 |
| 3 | Gemma 4 31B (IT) | 92 | #178 |
| 4 | Gemini 3 Flash (Preview) | 92 | #90 |
| 5 | GPT-5.6 Terra (Max) | 91 | #40 (GPT-5.6 Terra) |
| 6 | DeepSeek V4.1 Flash | 90 | #37 |
| 7 | Gemini 3.5 Flash | 89 | #66 |
| 8 | GPT-5.6 Sol (Max) | 89 | #22 (GPT-5.6 Sol) |
| 9 | Grok 4.5 (High) | 89 | #54 (Grok 4.5) |
| 10 | Claude Sonnet 4.6 (Thinking) | 89 | #94 (Claude Sonnet 4.6) |
| 11 | GPT-5.2 (2025-12-11) (xHigh) | 89 | |
| 12 | Qwen 3.6 Max (Preview) (Thinking) | 89 | |
| 13 | Gemini 3.6 Flash | 88 | #57 |
| 14 | DeepSeek V4 Flash (Max) | 88 | #155 (DeepSeek V4 Flash) |
| 15 | DeepSeek V4 Pro (0813) (Max) | 88 | #58 (DeepSeek V4 Pro (0813)) |
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-jcola-out-of-domain · How It Works · Data refreshed daily, snapshot 2026-11-05.