Finance Agent v2 - All Pass
Professional Work · 2026-05-19
Finance Agent v2 strict All-Pass grading metric over all 450 held-out test questions. Every required answer component must pass; published results are means over three model runs.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 47.7 |
| Gemini 3.5 Flash | 45.7 |
| Claude Fable 5 | 45.7 |
| Muse Spark 1.1 | 44.9 |
| Gemini 3.6 Flash | 44.6 |
| Opus 4.8 | 43.5 |
| GPT-5.6 Luna | 43.3 |
| Sonnet 5 | 42.1 |
| Kimi K3 | 41.7 |
| GPT-5.6 Sol | 41.4 |
| GPT-5.6 Terra | 40.4 |
| GPT-5.5 | 39.6 |
| Sonnet 4.6 | 38.8 |
| Opus 4.7 | 38.6 |
| GLM-5.2 | 38.0 |
| Grok 4.5 | 36.7 |
| MiniMax M3 | 36.7 |
| Gemini 3.5 Flash-Lite | 35.3 |
| Qwen3.7-Max | 34.6 |
| Inkling | 33.4 |
| GLM-5.1 | 32.7 |
| Kimi K2.6 | 32.6 |
| GPT-5.4 Mini | 32.4 |
| DeepSeek-V4-Pro | 31.4 |
| Gemini 3.1 Pro Preview | 30.9 |
| Gemini 3 Flash Preview | 30.2 |
| Qwen3.6 Plus (2026-04-02) | 29.4 |
| MiMo-V2.5-Pro | 28.8 |
| Grok 4.3 | 26.4 |
| GPT-5.4 Nano | 26.0 |
| Qwen3.7-Plus | 25.8 |
| Nemotron 3 Ultra 550B A55B | 24.8 |
| MiMo-V2.5 | 23.8 |
| Mistral Medium 3.5 | 21.6 |
| Haiku 4.5 | 19.8 |
| Gemini 3.1 Flash-Lite Preview | 19.3 |
| Grok 4.20 | 17.5 |
| MiniMax M2.7 | 17.1 |
| Laguna M.1 | 14.4 |
| Laguna XS.2 | 6.8 |