General Broad Information Seeking on WideSeekBench (test)
0.14Success Rate (Pass@4)Seed-1.8
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Seed-1.8Model Type=Proprietary2026.02 | 0.14 | 3.44 | 5.92 | 17.88 | 25.23 | — | — | |
| WideSeek-8B-SFTOptimization=SFT, Base Model=Qwen3-8B2026.02 | 0.14 | 1.74 | 3.66 | 11.35 | 18.92 | — | — | |
| GPT-5.2Model Type=Proprietary2026.02 | 0 | 4.45 | 6.75 | 21.03 | 26.88 | — | — | |
| GPT-5.1Model Type=Proprietary2026.02 | 0 | 4.11 | 6.75 | 20.44 | 27.88 | — | — | |
| DeepSeek-v3.2Model Type=Proprietary2026.02 | 0 | 4.34 | 6.85 | 20.51 | 27.09 | — | — | |
| Kimi-K2-ThinkingModel Type=Proprietary2026.02 | 0 | 3.17 | 5.86 | 17.48 | 25.19 | — | — | |
| Qwen3-8B-ThinkingModel Type=Open-Sourced2026.02 | 0 | 0.53 | 1.51 | 7.37 | 12.71 | — | — | |
| Qwen3-30B-A3B-ThinkingModel Type=Open-Sourced2026.02 | 0 | 1.26 | 3 | 10.11 | 16.51 | — | — | |
| WideSeek-8B-RLOptimization=RL from scratch, Base Model=Qwen3-8B2026.02 | 0 | 1.09 | 2.59 | 10.86 | 16.61 | — | — | |
| WideSeek-8B-SFT-RLOptimization=SFT + RL, Base Model=Qwen3-8B2026.02 | 0 | 1.95 | 3.88 | 12.87 | 19.73 | — | — |