Unverified70% confidenceOpinionTime unknown
DeepSeek V4 Pro相比Qwen3.6 27B Q4领先约6%，如果不是在推理数学项上意外翻车，领先幅度可能超过10%
1
Sources
70%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026
Sources
DeepSeek V4 vs Qwen3.6实测：8大类150场景深度对比评测
bilibili尚书省说书人
Related Entities
Related Claims
UnverifiedDeepSeek V4 Pro的浮点运算量降至V3.2版本的27%（降低3.7倍）73% similarUnverifiedQwen3.5的参数激活率为4.3%，而DeepSeek V3的671B模型激活率为5.5%71% similarUnverifiedQwen3-4B在多项推理类评测指标上已与DeepSeek V3或GPT-4o相差无几70% similarUnverifiedQwen3 27B在整体智能榜单上与DeepSeek V-Flash系列处于同一梯队，得分在52~53分区间69% similarUnverified根据Commander Code官方内部评测数据，DeepSeek V4 Pro在高难度工具依赖任务中10次有6次击败了Claude Opus68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/17941API
curl https://kongchang.com/api/v1/knowledge/claims/17941MCP
get_claim(id=17941)