CFMME - Seal Detection: leaderboard

CFMME (2026) is a Chinese financial multimodal evaluation set for vision-language models, built from financial documents, charts, seals, tables and exam questions. This board is the mAP in percent for locating company and personal seals with boxes; higher is better.

Metric: Seal detection mAP (%): boxes and categories (company or personal seal) on 756 document images on CFMME (6,052 Chinese financial multimodal instances over eight financial image types), zero-shot, mean of three runs, the non-reasoning model of each family; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GLM-4.5VNon-reasoning37.45#754
2Qwen 3 VL 235B A22B Instruct30.27#532
3Qwen 3 VL 8B Instruct28.27#1029
4InternVL3.5-241B-A28BNon-reasoning0.42#733
5GPT-4o0.3#751
6InternVL3.5-8BNon-reasoning0.19#1604
7Llama 3.2 11B Instruct0#5174
8Kimi-VL-A3B-Instruct0#2253
9ERNIE 4.5 VL 28B A3B0#2121

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=cfmme-seal-detection · How It Works · Data refreshed daily, snapshot 2026-11-21.