Strata

Run 125B Qwen models locally on 8GB+ NVIDIA GPUs with one click

Visit Strata ↗ link: rel="ugc noopener"

Strata is an inference engine enabling local deployment of Qwen3.8-Flash-Next (125B MoE) on consumer-grade NVIDIA GPUs with 8GB+ VRAM. Supports Windows and Linux with straightforward one-click installation. Provides OpenAI/Anthropic-compatible API endpoints on localhost, with optional image input capabilities for multimodal inference.