FCMBench-Video (English) - Classification: leaderboard

FCMBench-Video (2026) English-language handheld document videos from financial credit workflows, task classification: F1 (%) of the document types the model recognizes in a multi-document video against the ground-truth types. Zero-shot; higher is better.

Metric: Classification: F1 (%) of the document types the model recognizes in a multi-document video against the ground-truth types, on FCMBench-Video's en-US subset (795 composed videos of 8 English document types, 5,362 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: Estimated already saturated. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 3.5 27B91.53#380
2Gemini 3 Pro (Preview)90.98#123
3Qwen 3 VL 32B Instruct87.23#607
4Doubao-Seed-1.6-vision86.41#366
5Qwen 3 VL 8B Instruct82.82#1043
6Qwen3 Omni 30B A3B Instruct79#928
7InternVL3-8B76.6#1795
8Ovis2.5-9B71.02#1235
9Kimi-VL-A3B-Instruct69.25#2272

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=fcmbench-video-english-classification · How It Works · Data refreshed daily, snapshot 2026-11-22.