MortgageTax - Semantic Extraction
Professional Work · 2025-03-05
The Semantic Extraction split of MortgageTax. This child benchmark separates a source-reported subtask or subtrack from the parent aggregate so scores at different grains do not share one benchmark_slug.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 62.7 |
| Gemini 3.1 Pro Preview | 62.4 |
| Gemini 3 Pro Preview | 62.2 |
| Opus 4.7 | 61.8 |
| Gemini 3 Flash Preview | 61.7 |
| Gemini 3.5 Flash-Lite | 61.5 |
| Sonnet 5 | 61.4 |
| GPT-5.5 | 61.4 |
| Claude Fable 5 | 61.1 |
| Sonnet 4.5 | 61.1 |
| Grok 4.5 | 60.8 |
| Qwen3.6 27B | 60.8 |
| GPT-5.4 | 60.7 |
| Sonnet 4.6 | 60.5 |
| GPT-5.6 Terra | 60.3 |
| Gemini 2.5 Pro | 60.3 |
| Opus 4.6 | 60.1 |
| Opus 4.8 | 60.1 |
| Opus 4.5 | 59.9 |
| Qwen3.5-Flash | 59.8 |
| Gemini 2.5 Pro Experimental 03-25 | 59.7 |
| Claude 3.7 Sonnet | 59.6 |
| GPT-5.6 Sol | 59.5 |
| MiniMax M3 | 59.5 |
| Gemini 3.1 Flash-Lite Preview | 59.1 |
| GPT-5.6 Luna | 59.0 |
| Grok 4.3 | 59.0 |
| Opus 4.1 | 58.8 |
| Gemini 3.5 Flash | 58.7 |
| GPT-4.1 Mini | 58.7 |
| O4 Mini | 58.7 |
| Qwen3.5 Plus (2026-02-15) | 58.7 |
| Sonnet 4 | 58.4 |
| Gemini 3.6 Flash | 58.4 |
| GPT-5 Mini | 58.3 |
| Opus 4 | 58.2 |
| Kimi K2.5 | 58.1 |
| Muse Spark 1.1 | 57.8 |
| Qwen3.6 Plus (2026-04-02) | 57.8 |
| Gemini 2.5 Flash Preview 04-17 | 57.7 |