Expired50% confidenceBenchmarkExact time
Muse Glimmer scored a total of 36 points in real-world testing, ranking second to last among 6 models, 4 points lower than Qwen 3.6 27B.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
8/16/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 8/30/2026(expired)
Sources
实测Meta开源Muse Glimmer：偏科的Agent专用模型
bilibili碳基生物退役指南8/11/2026
Related Claims
UnverifiedMuse Glimmer在Tel3 Banking智能体测试上达到24%，超过Qwen 3.6 27B的17%76% similarUnverifiedMuse Glimmer在DeepSearch QA得分74.6，SWE-Bench Pro得分51.275% similarUnverifiedMuse-Glimmer在九大维度中赢了五个，其中数据提取领先约4.55分为最大差距项，工具/CLI领先3.5分74% similarUnverifiedMuse Glimmer 30B scored 51.2 on the SWE benchmark, meaning it can solve approximately half of real-world software engineering problems.73% similarUnverifiedMuse Glimmer模型在Arena.ai文本能力榜单上位列第24名，代码能力榜单上位列第26名72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/755349API
curl https://kongchang.com/api/v1/knowledge/claims/755349MCP
get_claim(id=755349)