ValerioDolci/capy-flash

Qwen3.8-Flash-Next (MoE) and Qwen3.8-27B on one or two 16 GB RTX 50-series GPUs: experts offloaded to CPU, two-GPU tensor parallel. 64 GB of RAM is enough for Flash-Next. Formerly ninfer-tp2.

8 best points · 2 comments · 1 Hacker News post · 140 GitHub stars · View on GitHub

Hacker News discussions

• Show HN: Capy-flash – a 125B MoE model on two 16 GB consumer GPUs (8 points, 2 comments, by Feyd, 2026-10-16)