Qwen3.8 Weights Releasedmodelscope.cn 31 pointsapitman1 month ago4 commentsSaveHideCopy link On HNComments−throwfaraway41mohttps://huggingface.co/Qwen/Qwen3.8-2.4T-A95B−13hours1moAnyone know if a MoE model that can run on a gpu with 16GB VRAM (like 3.6-35B-A3B) is in the pipeline as well, or only the dense 27B one?−petu1moWhoever runs their Twitter wouldn't confirm / promise anything beyond 27B, but didn't say no either. So maybe in a month, maybe not.−spwa41moNext up: https://modelscope.cn/models/Qwen/Qwen3.8-27BProbably easier to run for everyone.
Comments
https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B
Anyone know if a MoE model that can run on a gpu with 16GB VRAM (like 3.6-35B-A3B) is in the pipeline as well, or only the dense 27B one?
Whoever runs their Twitter wouldn't confirm / promise anything beyond 27B, but didn't say no either. So maybe in a month, maybe not.
Next up: https://modelscope.cn/models/Qwen/Qwen3.8-27B
Probably easier to run for everyone.