BloomBench - Arabic - Apply: leaderboard
BloomBench (Almieyar-Oryx series, QCRI, 2026) is a bilingual English and Arabic multiple-choice benchmark for vision-language models organized by the six levels of the revised Bloom taxonomy, scored as zero-shot accuracy in percent from the answer letter parsed from generated text (chance 25). This board is accuracy on the 499 Modern Standard Arabic items of the Apply level (applying formulas, scientific principles and basic logic to an image); higher is better.
Metric: Accuracy (%) on the 499 Arabic BloomBench items at the Bloom level Apply (applying formulas, principles and basic logic operations to the image), 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: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 27B | 84 | #1631 |
| 2 | Gemma 4 31B | 83 | #394 |
| 3 | Gemma 3 12B | 80 | #2000 |
| 4 | Gemma 4 26B A4B | 80 | #615 |
| 5 | GPT-4o Mini | 71 | #1623 |
| 6 | Qwen 2.5 VL 7B | 70 | #1787 |
| 7 | Qwen 2 VL 7B | 69 | #2261 |
| 8 | Gemma 3 4B | 64 | #5064 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=bloombench-arabic-apply · How It Works · Data refreshed daily, snapshot 2026-11-13.