SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer
Long Video Gen / World Model / World Action Model (Embodied) Infrastructure
cuDNN Frontend is NVIDIA's modern, open-source entry point to the cuDNN library and a growing collection of high-performance open-source kernels.
Pure Rust Inference Engine
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.
One command turns a DGX Spark or ASUS Ascent GX10 into a private AI server: seven measured, pinned AI lanes (text up to 1M tokens, images, video with sound), run from a local cockpit on a laptop or a phone. Nothing leaves the box.
Hand-written NVFP4 W4A16 CUDA kernels for Volta
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.
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.
GLM-5.2-NVFP4-REAP-469B serving on SM120 (4× RTX PRO 6000 Blackwell) — one-command vLLM launch recipe, 250K context, DeepSeek Sparse Attention + MTP speculative decode
AdaLLM is an NVFP4-first inference runtime for Ada Lovelace (RTX 4090) with FP8 KV cache and custom decode kernels. This repo targets NVFP4 weights and keeps the entire decode path in FP8
Bleeding-edge ComfyUI for NVIDIA DGX Spark (GB10/Blackwell/sm_121a). CUDA 13 + SageAttention v3 (sm_121a) + NVFP4 + 14 custom-node packs + Flux 2 Dev / LTX 2.3 22B / ACE-Step v1.5 XL Turbo pre-bundled with abliterated text-encoder paths.
Native Windows port of NInfer engine for RTX 5090. Features Qwen3.8-27B with QUASAR and NInfer models, MTP/DFlash2 with vision and 262,144 context
An environment where open models work as agents: a first-level harness with routing, skills, memory and sandboxed tools, on an engine that drafts and verifies its own tokens.
Serving 4-bit Qwen3.8-27B on a single DGX Spark (GB10): 75 tok/s single-stream, 246 tok/s aggregate at 8-way concurrency. NVFP4 vs MixedInt4-AutoRound vs the FP8 baseline, measured on one harness — including why the quantization advantage collapses to +0.2% by c16.
LLM server for one RTX 5090. Qwen3.8-27B at 91.0-93.7 tok/s (vLLM 68.9-70.0, llama.cpp 61.8-63.0). Runs Qwen3.8-Flash-Next (123.6 GiB) on a 32 GB card at 87.89-98.57 tok/s. Built for agents: tool calls, long context, many streams at once.
To associate your repository with the nvfp4 topic, visit your repo's landing page and select "manage topics."