~/bookmarks

William's Bookmark Library

/**/

GitHub - Niko1221/Strata: Qwen3.8-Flash-Next (125B MoE) on a 8GB+ NVIDIA GPU: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

github.comSaved September 28, 202610 min

Preview of GitHub - Niko1221/Strata: Qwen3.8-Flash-Next (125B MoE) on a 8GB+ NVIDIA GPU: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

Niko1221 · via GitHub

Summary

Strata runs the 125B-parameter Qwen3.8-Flash-Next MoE model on a single NVIDIA GPU (12-24 GB VRAM) with 64 GB of RAM, producing 52-90 tokens per second depending on quantization.

A one-click installer for Windows and Linux deploys the Strata inference engine, exposes OpenAI and Anthropic-compatible APIs on localhost, and supports optional image input. The project is free and open source, offering multiple model variants including a coding-specialized build that runs on 32 GB of RAM.

Topics

NVIDIA GPU Mixture-of-Experts Qwen3.8-Flash-Next Model Quantization Local LLM Inference

View on GitHub All Bookmarks

William's Bookmark Library

/**/

GitHub - Niko1221/Strata: Qwen3.8-Flash-Next (125B MoE) on a 8GB+ NVIDIA GPU: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

github.comSaved September 28, 202610 min

Niko1221 · via GitHub

Summary

Strata runs the 125B-parameter Qwen3.8-Flash-Next MoE model on a single NVIDIA GPU (12-24 GB VRAM) with 64 GB of RAM, producing 52-90 tokens per second depending on quantization.

A one-click installer for Windows and Linux deploys the Strata inference engine, exposes OpenAI and Anthropic-compatible APIs on localhost, and supports optional image input. The project is free and open source, offering multiple model variants including a coding-specialized build that runs on 32 GB of RAM.

Topics

NVIDIA GPU Mixture-of-Experts Qwen3.8-Flash-Next Model Quantization Local LLM Inference

View on GitHub All Bookmarks

~/bookmarks

GitHub - Niko1221/Strata: Qwen3.8-Flash-Next (125B MoE) on a 8GB+ NVIDIA GPU: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

Summary

Topics

Discover Similar Content

GitHub - julianmb/q38rocm: Qwen 3.8 27B ROCmFP4 on AMD Strix Halo (Ryzen AI Max+ 395). Up to 36 tok/s via MTP Speculation, TurboQuant & Mesa RADV Wave64.

GitHub - bilikaz/qwen38-flash-next-recipe: qwen38-flash-next recipe for solo GB10

GitHub - noonghunna/club-3090: Community recipes for serving LLMs on RTX 3090. Multi-engine (vLLM, llama.cpp, SGLang) and model-agnostic. Currently shipping Qwen3.6-27B configs for 1× and 2× cards.

Discover Similar Content

GitHub - Niko1221/Strata: Qwen3.8-Flash-Next (125B MoE) on a 8GB+ NVIDIA GPU: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

Summary

Topics

Discover Similar Content

GitHub - julianmb/q38rocm: Qwen 3.8 27B ROCmFP4 on AMD Strix Halo (Ryzen AI Max+ 395). Up to 36 tok/s via MTP Speculation, TurboQuant & Mesa RADV Wave64.

GitHub - bilikaz/qwen38-flash-next-recipe: qwen38-flash-next recipe for solo GB10

GitHub - noonghunna/club-3090: Community recipes for serving LLMs on RTX 3090. Multi-engine (vLLM, llama.cpp, SGLang) and model-agnostic. Currently shipping Qwen3.6-27B configs for 1× and 2× cards.