Leaderboard

Deep FinResearch Bench

Scores equity research reports from deep-research agents against professional analyst reports on three axes: qualitative rigour, forecast and valuation accuracy (SMAPE, target-price hit rate, recommendation direction), and how credible and verifiable each claim is.

Best reported

86

OpenAI

overall factual claim rate %

22 Apr 2026

Deep FinResearch Bench arXiv v1 (2026-04-22)

overall factual claim rate %, higher is better. 4 models.

Updated 22 Apr 2026

  1. OpenAI86
  2. Perplexity75.6
  3. Gemini69.6
  4. Grok53.2

Deep FinResearch Bench arXiv v1 (2026-04-22)

overall factual claim rate %, higher is better. 4 models.

Updated 22 Apr 2026

  1. 1OpenAI86
  2. 2Perplexity75.6
  3. 3Gemini69.6
  4. 4Grok53.2
Deep FinResearch Bench leaderboard (overall factual claim rate %, higher is better)
1OpenAIOpenAI86OpenAI DR (GPT-o3), Firm A-structured reportsPaper Table 7
Reported by the authors, 22 Apr 2026
2PerplexityPerplexity75.6Perplexity DR (Sonar Reasoning), Firm A-structured reportsPaper Table 7
Reported by the authors, 22 Apr 2026
3GeminiGoogle69.6Gemini DR (Gemini 2.5 Pro), Firm A-structured reportsPaper Table 7
Reported by the authors, 22 Apr 2026
4GrokxAI53.2Grok DR (Grok 4), Firm A-structured reportsPaper Table 7
Reported by the authors, 22 Apr 2026

Deep FinResearch Bench arXiv v1 (2026-04-22)

qualitative overall (1-4, GPT-5 judge), higher is better. 4 models.

Updated 22 Apr 2026

  1. Gemini2.3
  2. Perplexity2.1
  3. Grok2.0
  4. OpenAI2.0

Deep FinResearch Bench arXiv v1 (2026-04-22)

qualitative overall (1-4, GPT-5 judge), higher is better. 4 models.

Updated 22 Apr 2026

  1. 1Gemini2.3
  2. 2Perplexity2.1
  3. 3Grok2.0
  4. 3OpenAI2.0
Deep FinResearch Bench leaderboard (qualitative overall (1-4, GPT-5 judge), higher is better)
1GeminiGoogle2.3DefaultPaper Table 3
Reported by the authors, 22 Apr 2026
2PerplexityPerplexity2.1DefaultPaper Table 3
Reported by the authors, 22 Apr 2026
3GrokxAI2.0DefaultPaper Table 3
Reported by the authors, 22 Apr 2026
3OpenAIOpenAI2.0DefaultPaper Table 3
Reported by the authors, 22 Apr 2026

Full details for Deep FinResearch Bench