Mixed four-GPU rig on Strata: why 10GB of VRAM stayed unused

Critical-Entry3377Oct 7

Investigated why Strata 0.1.40.1 left VRAM free on a 5060 Ti, 3090 and two 3060s running Flash-Next IQ3 at 262k context, and found it was by design. A useful look at how the engine sizes its caches.

Open on Reddit

Launch: Qwen3.8-Flash-Next
Use case: Local inference
Format: Build
Source: reddit.com

Critical-Entry3377

See all

Niko1221

@Niko1221 · Sep 24

Open-source inference engine with an OpenAI/Anthropic-style local API that fits the 125B MoE onto a 12GB consumer GPU plus system RAM. The project behind the model's local-running surge.