EvoBrowseComp-ZH: leaderboard
The Chinese half of EvoBrowseComp (2026): 400 multi-hop questions built from fresh live-web knowledge that resists parametric recall, answered by search agents with web tools. Scored by LLM-judged accuracy; higher is better.
Metric: Accuracy (%) on the 400 Chinese EvoBrowseComp questions (complex multi-hop questions synthesized from fresh live-web knowledge to resist parametric recall), search agent with web search and page visit tools (Tongyi DeepResearch tool definitions), answers judged correct against the reference by a GLM-5-Chat judge; temperature 0.6, top_p 0.95, 128K context, at most 40 tool calls; higher is better. Source: arxiv.org. Saturation forecast: Around February 2028. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 (Max) | 38.3 | #56 (Claude Opus 4.8) |
| 2 | Claude Opus 4.6 (Max) | 36.8 | #102 (Claude Opus 4.6) |
| 3 | Qwen 3.5 397B A17B | 34.5 | #246 |
| 4 | GLM-5 | 30.5 | #260 |
| 5 | DeepSeek V3.2 (Thinking) | 30.5 | #406 (DeepSeek V3.2) |
| 6 | DeepSeek V4 Flash (High) | 24.8 | #258 (DeepSeek V4 Flash) |
| 7 | Qwen 3.5 122B A10B | 21 | #331 |
| 8 | DeepSeek V4 Flash (Non-reasoning) | 20 | #258 (DeepSeek V4 Flash) |
| 9 | Kimi K2.6 | 19.8 | #178 |
| 10 | Qwen 3.5 27B | 18.6 | #359 |
| 11 | Qwen 3 235B A22B 2507 (Thinking) | 17.5 | #463 |
| 12 | DeepSeek V4 Pro (Max) | 15 | #188 (DeepSeek V4 Pro) |
| 13 | Qwen 3.5 35B A3B | 14.2 | #482 |
| 14 | DeepSeek V4 Flash (Max) | 10.8 | #258 (DeepSeek V4 Flash) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=evobrowsecomp-zh · How It Works · Data refreshed daily, snapshot 2026-11-13.