FinMTM - Multiple-Choice: leaderboard
Metric: FinMTM multiple-choice score (%) on the 1,982 negative-selection multiple-choice questions (select every incorrect option) built from expert-validated financial chart and report questions (Chinese and English), set-overlap scoring with no over-selection: any wrong option scores 0, otherwise credit is the share of correct options chosen; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 79.6 | #91 |
| 2 | Gemini 3 Pro | 78.4 | #77 |
| 3 | Gemini 3 Flash | 78.1 | #93 |
| 4 | O3 | 73.3 | #121 |
| 5 | GLM-4.5V | 51 | #339 |
| 6 | Qwen 3 VL 30B A3B (Thinking) | 49.4 | #338 (Qwen 3 VL 30B A3B) |
| 7 | GPT-4o | 49.1 | #333 |
| 8 | Qwen 3 VL 235B A22B Instruct | 48.5 | #264 |
| 9 | Qwen 3 VL 30B A3B Instruct | 47.3 | #365 |
| 10 | Grok 4 Fast (Non-reasoning) | 46.8 | #242 (Grok 4 Fast) |
| 11 | Qwen 3 VL 32B (Thinking) | 46.5 | #287 (Qwen 3 VL 32B) |
| 12 | InternVL3-78B | 42.4 | #345 |
| 13 | Qwen 3 VL 235B A22B (Thinking) | 42.3 | #228 (Qwen 3 VL 235B A22B) |
| 14 | Qwen 3 VL 32B Instruct | 39.9 | #276 |
| 15 | Qwen 3 VL 4B Instruct | 34.2 | #506 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=finmtm-multiple-choice · How It Works · Data refreshed daily, snapshot 2026-10-11.