FCMBench V1.1 - Rationality Review: leaderboard
Evaluates visual credit-document understanding on FCMBench V1.1: 26 document types, 5,198 images and 13,806 VQA pairs. Exact-match key/value sets are scored with F1; task scores macro-average fine-grained subtasks. Numeric calculation permits differences below 2. The paper prefers instruct or lowest available reasoning effort. The two-stage DeepSeek-OCR plus V3.2 system stays distinct from either model alone. These are paper V1.1 scores, not the differing repository leaderboard. This series reports FCMBench V1.1 - Rationality Review using F1 (%).
Metric: F1 (%). Source: arxiv.org. Saturation forecast: Around February 2028. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 69.56 | #82 |
| 2 | Kimi K2.5 | 68.31 | #152 |
| 3 | Gemini 3 Flash | 62.82 | #103 |
| 4 | Claude Opus 4.5 | 62.02 | #87 |
| 5 | Qwen 3 VL 32B Instruct | 54.65 | #290 |
| 6 | Qwen 3 VL 30B A3B Instruct | 51.59 | #390 |
| 7 | Qwen 3 VL 8B Instruct | 51.24 | #419 |
| 8 | Qwen 3.5 122B A10B | 50.45 | #184 |
| 9 | Qwen 3 VL 235B A22B Instruct | 49.97 | #280 |
| 10 | GPT-5.2 | 49.1 | #115 |
| 11 | Qwen 3.5 35B A3B | 48.18 | #268 |
| 12 | Qwen 3.5 397B A17B | 45.8 | #154 |
| 13 | Phi-4 Multimodal Instruct | 45.8 | #922 |
| 14 | GLM-4.5V | 45.56 | #358 |
| 15 | GPT-5.1 | 43.98 | #145 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=fcmbench-v1-1-rationality-review · How It Works · Data refreshed daily, snapshot 2026-11-05.