List your company

Data Research Agent Evaluation on DRACO 50 tasks sample

69.36Normalized ScoreDeepSeek-V4-Flash
36.49645.02853.5662.092Jun 8, 2026

Evaluation results

MethodLinks
DeepSeek-V4-FlashSetting=RGI Turn 32026.0669.3674.5931.528.96
DeepSeek-V4-FlashSetting=RGI Turn 22026.0665.3571.139.6113.41
DeepSeek-V4-FlashSetting=Turn 12026.0657.263.94——
DeepSeek-V4-FlashSetting=SR Turn 22026.0656.6663.8426.1815.99
GPT-4.1Setting=RGI Turn 22026.0656.1962.2236.8816.87
GPT-4.1-miniSetting=RGI Turn 32026.0654.4560.9227.4618.59
GPT-4.1-miniSetting=RGI Turn 22026.0653.1159.9134.7814.52
GPT-4.1Setting=RGI Turn 32026.0651.2258.8627.1723.57
GPT-4.1Setting=SR Turn 22026.0644.8651.9415.5814.74
GPT-4.1Setting=Turn 12026.0644.7751.55——
GPT-4.1-miniSetting=SR Turn 22026.0640.1848.6415.412.9
GPT-4.1-miniSetting=Turn 12026.0637.7645.89——