FCMBench-Video (Chinese) - Cross-Document Validation: leaderboard
FCMBench-Video (2026) Chinese-language handheld document videos from financial credit workflows, task cross-document validation: exact-match accuracy (%) on consistency checks, numerical comparisons and business-rule decisions across documents, abstaining with Missing when evidence is absent. Zero-shot; higher is better.
Metric: Cross-Document Validation: exact-match accuracy (%) on consistency checks, numerical comparisons and business-rule decisions across documents, abstaining with Missing when evidence is absent, on FCMBench-Video's zh-CN subset (405 composed videos of 20 Chinese financial-credit document types, 5,960 instances); zero-shot with structured outputs; each model's native video serving path (2 FPS sampling for most open models, 0.5 FPS for Ovis2.5, raw video upload for the API models); higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 8 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 27B | 52.59 | #375 |
| 2 | Qwen 3 VL 8B Instruct | 48.15 | #1029 |
| 3 | Ovis2.5-9B | 45.56 | #1225 |
| 4 | Qwen3 Omni 30B A3B Instruct | 44.07 | #930 |
| 5 | Qwen 3 VL 32B Instruct | 40.74 | #606 |
| 6 | Gemini 3 Pro (Preview) | 38.29 | #119 |
| 7 | Doubao-Seed-1.6-vision | 33.33 | #361 |
| 8 | InternVL3-8B | 30.74 | #1782 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=fcmbench-video-chinese-cross-document-validation · How It Works · Data refreshed daily, snapshot 2026-11-21.