FreeToken Runs 290B MoE Models Locally on 8GB GPUs

FreeToken is an open-source edge-native MoE inference engine that claims to run frontier-scale 290B-parameter models locally on consumer hardware with as little as 8GB of VRAM. The project has already gained around 4.8k stars on GitHub.

2026-08-25 ~ 2026-08-25 · 2 related posts

• FreeToken Engine: Run 290B+ Frontier MoE Models Locally on 8GB Laptop GPUs — Saboo_Shubham_ · 2026-08-25
• FreeToken: run 290B+ frontier MoE models locally on your gaming PC — Saboo_Shubham_ · 2026-08-25