qwen3-8

Up to 4× faster LLM decoding on Apple Silicon, lossless. Native MLX port of DeepSeek's DSpark & z-lab's DFlash speculative decoding — Gemma-4, Qwen3.8, Muse-Glimmer, Nemotron, LFM2.5, Ornith-1.0, ternary Bonsai-27B.

Self-hosted Qwen3.8-27B (FP8) inference with vLLM, KServe and Envoy AI Gateway on RTX 6000 PRo or 2× RTX 4080 Super

Qwen3.8-27B on one DGX Spark GB10: DFlash 2 for 65 tok/s C1 and 120 C4; EAGLE 3/1/4 for ~191 C8, ~330 C16 and ~535 C32; native 262K, pinned SGLang.

Qwen 3.8 4B (Distilled): Run The Open Source King Locally! - Offline prompt engineering engine.

Qwen3.8-27B on one laptop CPU: up to 2.52 token/s, 8 GB tested, no runtime accuracy loss. Native OpenAI-compatible API with function tools; no GPU or Python. | 单颗笔记本 CPU 运行 Qwen3.8-27B：最快 2.52 token/s，最低 8 GB 内存可运行，推理加速不牺牲准确性。原生 OpenAI 兼容接口支持函数工具；无需 GPU 或 Python。

Native macOS control center and local AI agent gateway for Qwen3.8 on Apple Silicon — MLX, DFlash2, OpenAI Responses, Anthropic Messages, Claude Code, Codex, OpenCode and Grok Build.

Correctness-first Qwen3.8-27B NVFP4, TurboQuant K8V4, full-quality vision, and local agent deployment

Qwen 3.8 2B, 4B & 9B Distilled — Run Locally on 4GB GPU! Real-world multi-task edge showcase with Ollama & GGUF.

Run Qwen3.8 27B-parameter reasoning model with 128k context on a single RTX 3090 or 4090, locally, with an OpenAI-compatible API.

Qwen 3.8 is LIVE NOW! Can It Survive 3 Brutal Tests? (Qwen 3.8 Max Benchmarks) - Technical guide, 2.4T parameter specifications, token pricing, and Canvas execution test prompts.

Run Qwen3.8-27B (NVFP4 4-bit) locally on an RTX 5090 — one click on Windows 11, one command on Linux. Ships a coding agent, an OpenAI-compatible API, and three serving backends. Private, unmetered, offline.

Reproducible RTX 5090 dual-runtime 128K agent recipe for Qwen3.8-27B

AWQ Quantization of Qwen3.8 27B

Qwen3.8 Flash Next FP8 on 4x CMP 170HX: PP3 transformer stages with the 51B PLE table on GPU 3

Standalone eugr/spark-vllm-docker recipes for Qwen3.8-27B NVFP4 and official FP8 on one DGX Spark with dflash2

Run Qwen3.8-27B locally in Claude Code Desktop with vision, tools, native reasoning and Claude Code CLI support. Validated on an NVIDIA RTX 5090 32 GB.

Standalone eugr/spark-vllm-docker recipes for Qwen3.8-27B NVFP4 and official FP8 on one DGX Spark

Serves cyankiwi/Qwen3.8-27B-AWQ-INT4 (dense 27B vision-language, reasoning, apache-2.0, ~21GB INT4 weights) via vLLM on a single A100-80GB. via Modal

按任务自动调参 Qwen3.8 reasoning-budget 的守门人

Windows llama.cpp recipe for Qwen3.8-27B on 24 GiB RTX 4090 and RTX 3090 GPUs.