Leaderboard

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.

Best reported

76.6

GeminiGoogle

Final Score (0-100)

20 Oct 2025

FinResearchBench arXiv v3 (2025-10-20)

Final Score (0-100), higher is better. 8 models.

Updated 20 Oct 2025

  1. Gemini76.6
  2. OpenAI74.2
  3. Cashcat Pro70.7
  4. Minimax*67.7
  5. Reportify67.2
  6. Perplexity67.0
  7. Genspark*64.3
  8. AutoGLM60.0

FinResearchBench arXiv v3 (2025-10-20)

Final Score (0-100), higher is better. 8 models.

Updated 20 Oct 2025

  1. 1Gemini76.6
  2. 2OpenAI74.2
  3. 3Cashcat Pro70.7
  4. 4Minimax*67.7
  5. 5Reportify67.2
  6. 6Perplexity67.0
  7. 7Genspark*64.3
  8. 8AutoGLM60.0
FinResearchBench leaderboard (Final Score (0-100), higher is better)
1GeminiGoogle76.6all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025
2OpenAIOpenAI74.2all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025
3Cashcat ProNot stated70.7all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025
4Minimax*MiniMax67.7all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025
5ReportifyNot stated67.2all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025
6PerplexityPerplexity67.0all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025
7Genspark*Not stated64.3all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025
8AutoGLMZhipu AI60.0all 70 questionsPaper Table 1
Reported by the authors, 20 Oct 2025

Full details for FinResearchBench