Financial numerical reasoning on FinanceMath (val)
76AccuracyMOCA-AGENT
Evaluation Results
| Method | Links | |
|---|---|---|
| MOCA-AGENTBackbone=Qwen3.6-27B, Type=Open-source LLMs2026.06 | 76 | |
| GPT-4oType=Proprietary LLMs2026.06 | 67 | |
| Claude-3.5-SonnetType=Proprietary LLMs2026.06 | 60.6 |