Unverified85% confidenceFactTime unknown
QwQ-32B在多项主流基准测试中与DeepSeek R1满血版表现几乎持平，部分任务上实现了反超
1
Sources
85%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026
Sources
阿里QwQ-32B开源：32B参数如何媲美671B的DeepSeek R1
bilibiliAI大模型全栈
Related Entities
Related Claims
Unverified基准测试显示Qwen3-235B的表现追平甚至超越了DeepSeek R1，在数学、编码、推理等类别中均占优势77% similarUnverified在LiveBench等代码相关任务上，QwQ-32B超过DeepSeek R1约两个百分点76% similarUnverifiedQwen2.5-72B 在多项基准测试中接近 Llama-3.1-405B 的表现68% similarUnverifiedQwen3.6-27B各项基准测试成绩在同规模（20B-30B参数级别）模型中处于领先地位66% similarUnverifiedReddit社区的大规模基准测试基于BeeLlama.cpp v0.4.0，在Qwen 3.6 27B和Gemma 4 31B两款模型上累计跑了413组配置对比，其中Qwen模型238组、Gemma模型175组66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23192API
curl https://kongchang.com/api/v1/knowledge/claims/23192MCP
get_claim(id=23192)