deepseek-v4.1POST

Local deployment tip: run DeepSeek V4.1 Flash-1017 with the Maya-XS quant

A reply advises that with enough memory any quant runs, but smaller is faster when nothing fits fully in VRAM, recommending Maya-XS for DeepSeek V4.1 Flash-1017.

Peasant Smith

@PeasantSmithArchived 8 days agoOriginally posted 2026.10.24

Peasant Smith @PeasantSmithOriginally posted 2026.10.24

Open original on X

With that amount of memory you can run any quant available really, however the smaller the faster you will run since neither can fit fully in VRAM. If you want GLM 5.3 Flash you can go with Maya-S24 of the smallest Exl3, if you want DeepSeek V4.1 Flash-1017 you can go with Maya-XS Let me know your results :)

x.com/peasantsmith/status/2113736026682253405Archived 2026.10.24 12:07