Publisher-reported results
FrontierFinance
Evaluate AI systems on open-ended investment research.
Rubric qualification rate
Higher is betterSource-reported top five of 18 systems; snapshot dated September 28, 2026.
| Rank | System / model | Score |
|---|---|---|
| 1 | Samaya System (high effort) | 56.0% |
| 2 | Claude Fable 5.1 · Finance Agent v2 | 55.9% |
| 3 | Samaya System | 52.9% |
| 4 | Gemini 3.8 Flash · Finance Agent v2 | 50.0% |
| 5 | Claude Fable 5 · Finance Agent v2 | 49.2% |
Source-reported snapshot. Stubbi has not rerun these evaluations.
Coverage
220 questions, 11,543 rubric items, six investment workflows; 18 systems in the source table.
Methodology
Overall rubric qualification rate. Systems use Samaya, Finance Agent v2, or web-search harnesses. Finance Agent v2 and Samaya share limits of 200 tool calls and a 300-second tool-call time limit. Thinking is enabled at default levels: high for Claude Fable 5.1 and Claude Fable 5; medium for Gemini 3.8 Flash.