← sources

Qwen3.8-Flash-Next @ 1.66bpw-exl3 STABLE under 36gb

source: reddit (reddit:LocalLLaMA)published 2026-10-21 07:21ingested 2026-10-22 02:43

retained text

Fits entirely in a 5090, no offloaded experts. Super-solid sub-2bpw!  
([Liquid cat demo video](https://x.com/turboderp_/status/2112695266767724627) generated by the quantized model)

From the Cat's mouth:

>Stable inference throughout, no reasoning loops, just smooth sailing at 200-300 t/s (and a lot of compacting since one 32 GB GPU still doesn't leave a lot of room for context, even with cache quantization.

[https://huggingface.co/turboderp/Qwen3.8-Flash-Next-exl3](https://huggingface.co/turboderp/Qwen3.8-Flash-Next-exl3)   

document fetched 2026-10-22 00:45 · canonical https://reddit.com/r/LocalLLaMA/comments/1xbgi6i/qwen38flashnext_166bpwexl3_stable_under_36gb