Financial Reasoning on OPENFIN (held-out)
73.4DBSEXPERT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| EXPERTModel=Qwen3-1.7B optimized2026.02 | 73.4 | — | |
| DataChef-32BOracle Upper Bound=true, Samples=322026.02 | 67.1 | — | |
| Qwen3-Next ⊕ Kimi-K2Reasoning backbone=Qwen3-Next-80B, Coding backbone=Kimi-K2-Instruct2026.02 | 64 | 54.7 | |
| DataChef-32B2026.02 | 63.9 | 67 | |
| SOURCEbest2026.02 | 63.7 | — | |
| Gemini-3-Pro2026.02 | 61.8 | 54.9 | |
| Kimi-K22026.02 | 46.5 | 51.5 | |
| SOURCEavg2026.02 | 41.7 | — | |
| Qwen3-32B2026.02 | 23.8 | 34.9 |