MohammadHaishemKhawaja/hashyy

A full, unpruned 177B MoE model at 11.5 tok/s on one 12 GB GPU. Expert streaming for llama.cpp, with the measurement harness and every dead end.

1 best points · 0 comments · 1 Hacker News post · 25 GitHub stars · View on GitHub

Hacker News discussions

• Qwen3.8-Flash-Next 177B running at 11–15 tok/s on 1 RTX 5070 12GB and 32GB RAM (1 points, 0 comments, by maziianobeatz58, 2026-10-04)