Nejumi 4 - jaster (0-shot) - AIO: leaderboard
AIO (AI King), Japanese open-domain quiz question answering, from the llm-jp-eval jaster suite as run 0-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 | Gemini 3 Flash (Preview) | 96.27 | #90 |
| 2 | Gemini 3.6 Flash | 95.86 | #57 |
| 3 | Gemini 3 Pro (Preview) | 95.62 | #71 |
| 4 | Grok 4 | 95.55 | #189 |
| 5 | Claude Opus 4 (Non-reasoning) | 95.19 | #170 (Claude Opus 4) |
| 6 | GPT-5.4 (2026-03-05) (xHigh) | 94.62 | #63 (GPT-5.4 (2026-03-05)) |
| 7 | Gemini 3.5 Flash | 94.55 | #66 |
| 8 | Muse Spark Open (xHigh) | 94.17 | #42 (Muse Spark Open) |
| 9 | DeepSeek V4 Pro (Max) | 94.16 | #112 (DeepSeek V4 Pro) |
| 10 | Claude Opus 4.7 (xHigh) | 93.86 | #51 (Claude Opus 4.7) |
| 11 | Gemini 3.1 Pro (Preview) | 93.68 | #64 |
| 12 | Gemini 2.5 Flash | 93.68 | #255 |
| 13 | Claude Opus 4.1 (Thinking) | 93.66 | #146 (Claude Opus 4.1) |
| 14 | Claude Opus 4.5 (Thinking) | 93.56 | #87 (Claude Opus 4.5) |
| 15 | GLM-5.2 (Max) | 93.47 | #80 (GLM-5.2) |
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-aio · How It Works · Data refreshed daily, snapshot 2026-11-05.