BloomBench - English - Create: leaderboard

Metric: Accuracy (%) on the 685 English BloomBench items at the Bloom level Create (choosing the synthesis that best meets creative constraints), 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 December 2026. 8 models tracked.

Top models

#ModelScore
1Gemma 4 31B86
2Gemma 4 26B A4B85
3Gemma 3 27B84
4Gemma 3 12B80
5Qwen 2.5 VL 7B79
6GPT-4o Mini77
7Qwen 2 VL 7B75
8Gemma 3 4B74

Interactive version: theaggregate.ai/benchmark?slug=bloombench-english-create · How It Works · Data refreshed daily, snapshot 2026-09-29.