Expired65% confidenceFactTime unknown
RTX 5090上Qwen3.6 27B模型预填充速度从174提升到253 token/s，显存占用约18GB
3
Sources
65%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026(expired)
Sources
Qwen3.6 MTP加速实测：单GPU推理飙到220 token/s
bilibili普通鱼学家呀
Related Entities
Related Claims
UnverifiedRTX 5090上Qwen3.6 35B-A3B推理速度达220 token/s，显存占用约23GB90% similarUnverified在显存充足、模型完整驻留GPU时，RTX 4090跑Q4量化的27B模型通常可达15～30 tokens/s；触发CPU卸载后速度可能骤降至1～2 tokens/s甚至更低81% similarUnverified在24GB显存的显卡上运行Qwen3.6 27B模型，未经优化可达每秒约40个Token的生成速度，优化后可达50-60 Token/s80% similarVerified该工作流在RTX 3060 6GB显存加16GB内存配置下即可运行79% similarUnverified对于RTX 4090（24GB显存），WhichLLM可能会推荐Qwen3.6 27B的Q5量化版本，预计推理速度约为每秒27 Token78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/18059API
curl https://kongchang.com/api/v1/knowledge/claims/18059MCP
get_claim(id=18059)