Unverified60% confidenceBenchmarkExact time
Muse Glimmer 30B is reported to achieve 74.9 tokens per second on an RTX 5090, and reviewers observed approximately 60–65 tokens/sec running the full-precision version on an RTX 3090.
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
8/17/2026
First Seen
Valid until: 11/15/2026
Sources
Meta Muse Glimmer 30B深度评测：视觉理解能力惊艳，本地部署门槛低
Related Claims
UnverifiedMuse-Glimmer-30B在智能体工作流得90.87分、指令遵循90.86分、工具/CLI 88.84分，三项均优于Corpus 3.6 27B76% similarUnverifiedMuse Glimmer 30B scored 51.2 on the SWE benchmark, meaning it can solve approximately half of real-world software engineering problems.70% similarUnverifiedThe 7900XTX on Windows achieves only approximately 400–479 t/s for the prompt processing (preview) phase with Qwen3 IQ3_XSS quantization.70% similarUnverifiedMuse Glimmer在Tel3 Banking智能体测试上达到24%，超过Qwen 3.6 27B的17%69% similarUnverifiedThe Qwen3.8-27B checkpoint ships with a multi-token prediction (MTP) draft head, containing 22 tensors and 0.44 GiB, located in the same download package, and speculative decoding can be enabled via a launch parameter.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/762694API
curl https://kongchang.com/api/v1/knowledge/claims/762694MCP
get_claim(id=762694)