Finance Agent
Can the model produce useful work in a professional knowledge-work setting?
Knowledge Work8 ranked models10 reported values2 reportsHigher is better
Top rankings
One row per model, using its best reported score across effort settings.
| Rank | Model | Best score | Best reported setting |
|---|---|---|---|
| 1 | Gemini 3.5 FlashGoogle | 57.9% | Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source |
| 2 | Muse Spark 1.1Meta | 57.2% | xhigh effortMeta report1 value · 1 reportJul 9, 2026 · source |
| 3 | Claude Opus 4.8Anthropic | 53.9% | max effortMeta report1 value · 1 reportJul 9, 2026 · source |
| 4 | GPT-5.5OpenAI | 51.8% | xhigh effortMeta report2 values · 2 reportsJul 9, 2026 · source |
| 5 | Claude Opus 4.7Anthropic | 51.5% | Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source |
| 6 | Claude Sonnet 4.6Anthropic | 51.0% | Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source |
| 7 | Gemini 3.1 ProGoogle | 43.0% | high effortMeta report2 values · 2 reportsJul 9, 2026 · source |
| 8 | Gemini 3 FlashGoogle | 42.6% | Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source |
Effort curve
Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.
No cost-linked effort sweep for this version.
Definition and comparison boundarypublic methodology
Can the model produce useful work in a professional knowledge-work setting?
Professional analysis, document, finance, office, or domain-specific tasks. Higher is better. The value is the percentage reported in this lab's table.
Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.
927 expert-reviewed questions; values sourced from the cited evaluator.
Method / source