nvfp4

Open repos and models their makers tagged nvfp4, most starred first. The same for everyone.

AI on my own machineRepo

auto-round

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

  • 1.7k
  • Python
  • Apache-2.0
  • 2 h ago
intelTake it
AI on my own machineRepo

Qwen3.6-27B-AEON-Ultimate-Uncensored-DFlash

Fully uncensored, capability-enhanced abliteration of Qwen3.6-27B. NVFP4 + z-lab DFlash speculative decoding (n=12) on the unified ghcr.io/aeon-7/aeon-vllm-ultimate:latest container, tuned for long-context draft acceptance on DGX Spark. 6 HF variants (BF16/NVFP4/MTP/MTP-XS), docker-compose, and QuickStart.

  • 498
  • Python
  • Apache-2.0
  • 6 d ago
AEON-7Take it
Run my own servicesRepo

alzeo

Alzeo is an open-source installer that turns an NVIDIA DGX Spark, or any GB10 machine (ASUS, Dell, HP, Lenovo, MSI, GIGABYTE, Acer), into a private AI server in one command: measured model lanes, OpenAI and Anthropic APIs, and a local cockpit.

  • 269
  • Python
  • MIT
  • 4 h ago
hasso5703Take it
GitHubRepo

strata-nvfp4

Strata fork: Qwen3.8-Flash-Next 125B MoE in NVFP4 on one RTX 20-50 card (12 GB+) and 64 GB of RAM or more. Setup installs our GPTQ quants (huihui abliterated, OrcaRouter) from Hugging Face. CPU experts on AVX-512 or AVX2 (+AVX-VNNI), all experts in RAM from 92 GB, a low-RAM mode below, vision, W4A8 prompts. Windows / CUDA 13.

  • 249
  • C++
  • MIT
  • 1 d ago
sergqwerTake it
AI on my own machineRepo

sglang-rtxpro6000

Optimized SGLang runtime for Qwen3.8-27B FP8 with DFlash2 and Qwen3.8 Flash-Next NVFP4 with FR-Spec on one NVIDIA RTX PRO 6000 Blackwell 96 GB GPU (SM120): 524K context, HiCache and NIXL.

  • 203
  • Python
  • Apache-2.0
  • 1 d ago
jpezzulliTake it
Run my own servicesRepo

imp

Run large AI chat models privately on your own NVIDIA RTX 50 or RTX PRO graphics card. On one RTX 5090, Qwen3.8-27B writes about 70 words per second, 1.4x faster than vLLM. Even a model almost four times larger than the card's memory runs. Works with apps made for ChatGPT or Claude. Starts with one docker command.

  • 83
  • C++
  • MIT
  • 6 h ago
kekzlTake it
AI on my own machineRepo

ignis

An inference engine for one NVIDIA RTX card, built for a coding agent and its subagents, and a companion that listens, reads and draws: Qwen3.8 27B and 125B Flash-Next MoE, Qwen-Image-2.1, Qwen3-ASR on the CPU. OpenAI-compatible API and Playground in one binary.

  • 67
  • Rust
  • Apache-2.0
  • 8 h ago
gpillonTake it
AI on my own machineRepo

llmash

Ollama-compatible LLM server. Beats vllm by 40% on single stream decode, and by 20% on multi stream decode, and beats ollama by much 2-4x. Made possible with custom CUDA kernels, MTP, speculation and its own 3-bit builds. Uncensors models in one command. Windows and Linux, NVIDIA. Optimized to not eat up ram.

  • 37
  • C++
  • Apache-2.0
  • 6 d ago
omgitsbaseTake it
Images and videoRepo

minimax-h3-tensorfold-rtx

MiniMax H3 audio-video model on TensorFold NVFP4 kernels: a faster ComfyUI loader for one RTX 50-series GPU (2.3x stock at 20 steps, 81 s per 5 s 1344x768 video with Turbo + sparse attention on an RTX 5070 Ti)

  • 10
  • Python
  • Apache-2.0
  • 3 wk ago
jayleatonTake it
Search GitHub and Hugging Face for nvfp4

Get , open it once and sign in. Then come back here and press the button again.