Unverified50% confidenceBenchmarkExact time
Corpus 3.6 27B在深度任务、结构化输出、编程能力上胜出，其中编程能力领先1.66分
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/16/2026
First Seen
Valid until: 11/14/2026
Sources
Muse-Glimmer-30B实测：24G显存跑出九维跑分76分的性价比之王
Related Claims
Unverified千问3.8 27B的代码方向进步明显大于文本创作，剧本拆解能力相比3.6有提升但幅度有限67% similarUnverifiedCorpus 3.6 27B是开源社区基于阿里千问3.6（Qwen-3.6）进行后训练得到的优化版本，继承了千问3.6的256K长上下文能力65% similarUnverifiedAt a draft length of 15 tokens, DSpark's efficiency gains in math, code, and dialogue scenarios reached 30%, 26%, and 22% respectively.64% similarUnverified通义千问3.6 27B在语义三次全对率和格式正确率上均领先Muse Glimmer59% similarUnverifiedbase64 编码会引入约 33% 的体积膨胀，膨胀比为 8/6≈1.333，因为每个字节只携带6位有效信息56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/757567API
curl https://kongchang.com/api/v1/knowledge/claims/757567MCP
get_claim(id=757567)