# FinResearchBench

Measures how well deep-research agents write financial research outputs (stock, event, sector, morning brief, post-market recap and similar). An Agent-as-a-Judge first extracts a logic tree of arguments and evidence, then scores the output against it.

## Facts

- Maker: Stepfun; FinStep
- Year: 2025
- Use case: Analyst Financial research
- Size: 70 questions (10 for each of 7 task types)
- Public split: none (no data or code linked)
- Grading: Ten dimensions, each scored 0-100 and weighted equally into a Final Score. Six are scored by an LLM with a rubric (Claude 3.7 Sonnet is the core judge) and four are computed by rules.
- Licence: Not stated
- Submission: none
- Products allowed: Yes
- Latest version: arXiv v3 (2025-10-20)
- What it does not cover: Does not check numeric facts against sources or forecast accuracy. It scores structure and argument quality.
- Last checked: 1 Nov 2026
- Links: paper https://arxiv.org/abs/2507.16248
- Directory page: https://benchmarks.finance/benchmarks/finresearchbench/
- Leaderboard: https://benchmarks.finance/leaderboards/finresearchbench/

## Results

### FinResearchBench: Final Score (0-100) (higher is better, 8 models)

| Rank | Model | Provider | Score | Metric | Setting | Reported by | Source | URL | Date |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | Gemini | Google | 76.6 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |
| 2 | OpenAI | OpenAI | 74.2 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |
| 3 | Cashcat Pro | Not stated | 70.7 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |
| 4 | Minimax* | MiniMax | 67.7 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |
| 5 | Reportify | Not stated | 67.2 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |
| 6 | Perplexity | Perplexity | 67.0 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |
| 7 | Genspark* | Not stated | 64.3 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |
| 8 | AutoGLM | Zhipu AI | 60.0 | Final Score (0-100) | all 70 questions | Reported by the authors | Paper Table 1 | https://arxiv.org/html/2507.16248v3 | 20 Oct 2025 (checked 1 Nov 2026) |


Source: https://benchmarks.finance/benchmarks/finresearchbench/

Created and maintained by Moneda.