FinMTM - Financial Agent (Fuzzed): leaderboard
Metric: FinMTM financial agent score (0-100) over 1,000 tool-use tasks with MCP financial tools whose queries obfuscate entity names through knowledge-base substitutions and masked image regions: tool-call quality (recall-weighted F2 of predicted against reference calls, up to 25) plus judged reasoning (up to 25) and answer correctness against the ground truth (up to 50); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 53.6 | #93 |
| 2 | Gemini 3 Pro | 48.3 | #77 |
| 3 | GPT-5 | 35.9 | #91 |
| 4 | Qwen 3 VL 235B A22B (Thinking) | 35.2 | #228 (Qwen 3 VL 235B A22B) |
| 5 | Qwen 3 VL 235B A22B Instruct | 32.1 | #264 |
| 6 | O3 | 31.4 | #121 |
| 7 | Grok 4 Fast (Non-reasoning) | 30.2 | #242 (Grok 4 Fast) |
| 8 | GPT-4o | 29.7 | #333 |
| 9 | GLM-4.5V | 26.5 | #339 |
| 10 | Qwen 3 VL 32B (Thinking) | 23.2 | #287 (Qwen 3 VL 32B) |
| 11 | Qwen 3 VL 32B Instruct | 19.6 | #276 |
| 12 | Qwen 3 VL 30B A3B (Thinking) | 18.9 | #338 (Qwen 3 VL 30B A3B) |
| 13 | InternVL3-78B | 18.2 | #345 |
| 14 | Qwen 3 VL 30B A3B Instruct | 16.2 | #365 |
| 15 | Qwen 3 VL 4B Instruct | 15.1 | #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-financial-agent-fuzzed · How It Works · Data refreshed daily, snapshot 2026-10-11.