CFMME - Formula Recognition: 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 CDM character detection matching score in percent for recognized formulas; higher is better.

Metric: Formula recognition CDM (%, character detection matching of the rendered formula) 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: Estimated already saturated. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Qwen 3 VL 235B A22B Instruct95.01#532
2InternVL3.5-8BNon-reasoning89.02#1604
3InternVL3.5-241B-A28BNon-reasoning88.06#733
4Qwen 3 VL 8B Instruct87.57#1029
5GPT-4o83.73#751
6GLM-4.5VNon-reasoning83.24#754
7Kimi-VL-A3B-Instruct71.44#2253
8ERNIE 4.5 VL 28B A3B70.98#2121
9Llama 3.2 11B Instruct47.23#5174

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

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