Deep Research on DeepResearch Bench 1.0 (test)
46.45Overall ScoreOpenAI-DeepResearch
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| OpenAI-DeepResearchProprietary=true2026.01 | 46.45 | 46.46 | 43.73 | 49.39 | 47.22 | |
| Kimi-ResearcherProprietary=true2026.01 | 44.64 | 44.96 | 41.97 | 47.14 | 45.59 | |
| DeepDive-30B + C-GRPOBackbone=DeepDive-30B, RL_Algorithm=C-GRPO2026.01 | 41.99 | 39.75 | 35.87 | 48.51 | 46.63 | |
| Tongyi-DeepResearchProprietary=true2026.01 | 40.46 | 39.46 | 34.44 | 46.22 | 44.27 | |
| DeepDive-30B + GRPOBackbone=DeepDive-30B, RL_Algorithm=GRPO2026.01 | 39.3 | 36.1 | 31.66 | 47.65 | 44.92 | |
| Grok-Deeper-SearchProprietary=true2026.01 | 38.22 | 36.08 | 30.89 | 46.59 | 42.17 | |
| DeepDive-4B + C-GRPOBackbone=DeepDive-4B, RL_Algorithm=C-GRPO2026.01 | 37.51 | 33.88 | 30.01 | 45.72 | 43.82 | |
| DeepDive-30B-SFTBackbone=DeepDive-30B, RL_Algorithm=SFT2026.01 | 37.51 | 34.27 | 28.85 | 46.77 | 43.21 | |
| DeepDive-4B + E-GRPOBackbone=DeepDive-4B, RL_Algorithm=E-GRPO2026.01 | 36.59 | 33.2 | 28.3 | 45.58 | 42.67 | |
| DeepDive-30B + E-GRPOBackbone=DeepDive-30B, RL_Algorithm=E-GRPO2026.01 | 36.12 | 32.31 | 27.73 | 45.72 | 42.33 | |
| DeepDive-4B + GRPOBackbone=DeepDive-4B, RL_Algorithm=GRPO2026.01 | 34.79 | 31.29 | 26.79 | 43.81 | 41.58 | |
| DeepDive-4B-SFTBackbone=DeepDive-4B, RL_Algorithm=SFT2026.01 | 33.81 | 29.57 | 24.23 | 44.05 | 41.02 |