KVBench (Brief Caption) - Geography (Chinese): leaderboard

KVBench (2026) tests whether text-to-image models can draw correct senior high-school geography content, such as diagrams and scientific illustrations, from Chinese prompts given as a brief caption, so the model must supply the textbook knowledge itself. Qwen2.5-VL-32B checks each image against a list of textbook-grounded constraints; the score is the percentage satisfied, higher is better.

Metric: Checklist score (%) on the geography prompts in Chinese, each given as a brief caption, so the model must supply the textbook knowledge itself: KVBench knowledge-intensive text-to-image prompts from over 30 senior high-school textbooks (150 per subject and language, each with a textbook reference image); Qwen2.5-VL-32B-Instruct answers a checklist of four to six binary questions (key objects, attributes, spatial relations, reasoning outcomes) about each generated image, and the score is the mean share of satisfied items, in percent; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 14 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Nano Banana Pro (Gemini 3 Pro Image)63.29
2Seedream 4.052.29
3FLUX.2-dev49.62
4GPT-Image (KVBench checkpoint unspecified)44.31
5FLUX 2 Max31.96
6Qwen-Image29
7BAGEL-7B-MoT27.91#4619
8Z-Image27.33
9OmniGen227.13#4579
10FLUX.1-dev25.51
11Show-o222.71
12SD3.5-Large (KVBench GGUF build, quantization unspecified)17.78
13Janus-Pro12.53
14FLUX.1-Krea-dev10.67

Interactive version: theaggregate.ai/benchmark?slug=kvbench-brief-caption-geography-chinese · How It Works · Data refreshed daily, snapshot 2026-11-21.