Frontier Financial Judgement (atomic)
Finance & Accounting · 2026-07-22
Frontier Financial Judgement accuracy over its 246 individual newness, importance and direction labels, scored separately.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.5 | 71.1 |
| GPT-5.6 Sol | 69.1 |
| GPT-5.6 Terra | 63.8 |
| GPT-5.6 Luna | 63.4 |
| Sonnet 4.6 | 63.0 |
| Opus 4.8 | 59.8 |
| GPT-5.4 | 59.3 |
| DeepSeek-V4-Flash | 58.1 |
| DeepSeek-V4-Pro | 57.7 |
| MiMo-V2.5 | 56.9 |
| GPT-5.4 Mini | 54.1 |
| Qwen3.7-Plus (source unspecified) | 49.2 |
| Qwen3.6 35B A3B | 43.9 |
| Nemotron 3 Super 120B A12B | 15.9 |