BloomBench - Arabic - Evaluate: leaderboard
Metric: Accuracy (%) on the 592 Arabic BloomBench items at the Bloom level Evaluate (judging coherence, harm and safety, and image quality), four-option image-question multiple choice, zero-shot, temperature 0, the chosen letter parsed from the generated answer (Regex-based Answer Extraction; unparseable answers count as wrong); chance 25; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 4 31B | 93 |
| 2 | Gemma 3 27B | 89 |
| 3 | Gemma 4 26B A4B | 89 |
| 4 | Gemma 3 12B | 88 |
| 5 | GPT-4o Mini | 83 |
| 6 | Qwen 2.5 VL 7B | 82 |
| 7 | Qwen 2 VL 7B | 79 |
| 8 | Gemma 3 4B | 73 |
Interactive version: theaggregate.ai/benchmark?slug=bloombench-arabic-evaluate · How It Works · Data refreshed daily, snapshot 2026-09-29.