Nejumi 4 - jaster (2-shot) - NIILC: leaderboard
NIILC, Japanese encyclopedic question answering, 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: Around October 2028. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 64.42 | #64 |
| 2 | Muse Spark Open (xHigh) | 61.58 | #42 (Muse Spark Open) |
| 3 | GPT-5.6 Terra (Max) | 61.45 | #40 (GPT-5.6 Terra) |
| 4 | Claude Opus 4.7 (xHigh) | 61.27 | #51 (Claude Opus 4.7) |
| 5 | DeepSeek V4 Pro (0813) (Max) | 61.11 | #58 (DeepSeek V4 Pro (0813)) |
| 6 | DeepSeek V4 Pro (Max) | 60.83 | #112 (DeepSeek V4 Pro) |
| 7 | GPT-5.4 (2026-03-05) (xHigh) | 60.66 | #63 (GPT-5.4 (2026-03-05)) |
| 8 | Gemini 3 Pro (Preview) | 60.64 | #71 |
| 9 | Claude Opus 4.1 (Thinking) | 60.63 | #146 (Claude Opus 4.1) |
| 10 | Claude Opus 4.5 (Thinking) | 60.55 | #87 (Claude Opus 4.5) |
| 11 | Gemini 3.5 Flash | 60.5 | #66 |
| 12 | Qwen 3.6 Max (Preview) (Thinking) | 60.33 | |
| 13 | Gemini 3.6 Flash | 60.19 | #57 |
| 14 | GPT-5.6 Sol (Max) | 60.15 | #22 (GPT-5.6 Sol) |
| 15 | GPT-5.2 (2025-12-11) (xHigh) | 59.87 |
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-niilc · How It Works · Data refreshed daily, snapshot 2026-11-05.