Textual long-horizon complex reasoning on BrowseComp ZH
82.4ScoreDoubao-2.0-Pro
Evaluation results
| Method | Links | |
|---|---|---|
| Doubao-2.0-ProSize=-, Train=-2026.06 | 82.4 | |
| GPT-5.2Size=-, Train=-2026.06 | 76.1 | |
| MiroThinker-1.7Size=235B, Train=Mid/SFT/RL2026.06 | 75.3 | |
| GLM-5Size=744B, Train=-2026.06 | 72.7 | |
| MiroThinker-1.7-miniSize=30B, Train=Mid/SFT/RL2026.06 | 72.3 | |
| Qwen3.5-397BSize=397B, Train=-2026.06 | 70.3 | |
| DeepSeek-V3.2Size=671B, Train=-2026.06 | 65 | |
| Gemini-3.1-Pro-PreviewSize=-, Train=-2026.06 | 63.3 | |
| Kimi-K2.5Size=1T, Train=-2026.06 | 62.3 | |
| MiniMax-M2.7Size=230B, Train=-2026.06 | 56.4 | |
| REDSearcherSize=30B, Train=Mid/SFT/RL2026.06 | 49.8 | |
| Claude-4.6-Sonnet-ThinkingSize=-, Train=-2026.06 | 49.4 | |
| OpenSeeker-v1Size=30B, Train=SFT2026.06 | 48.4 | |
| S1-DeepResearchSize=32B, Train=SFT2026.06 | 48.4 | |
| Tongyi-DeepResearchSize=30B, Train=Mid/SFT/RL2026.06 | 46.7 | |
| Qwen3-235BSize=235B, Train=-2026.06 | 13.5 | |
| Qwen3-32BSize=32B, Train=-2026.06 | 7.3 |