Havuz

Qwen3.8-Flash-Next (125B MoE) up to 1M context on one or two RTX cards (tuned on 3080 10gb). A measured, tuned and benchmarked fork of Str

Локальная студия генерации кинематографичных видео со звуком на MiniMax H3 — Windows, NVIDIA GPU, полностью offline.

System monitor for workstations, laptops and NAS — terminal UI and native GUI, many hosts over SSH. A btop alternative.

Windows utility focused on cuda setup, nvidia local AI, and cuda environment for local AI workflows.

P104-100 (GP104) PCIe Gen2 x4 unlock tool for Windows. Gen1 x4 -> Gen2 x4: 0.78 -> 1.55 GB/s measured.

A decision library for running local LLMs on multi-GPU machines without NVLink: parallelism, communication, quantization, speculative decodi

A code reviewer that reports only what it can prove: a test that shows the bug, or the data that confirms the risk. Open source, built by a

Qwen3.8-Flash-Next (180B MoE) on one 12-16 GB NVIDIA GPU, Linux or Windows. Own engine, own compression.

Berth is a lightweight GPU resource manager written in Go. It controls which applications get access to your GPU, when they get it, and how

面向 DeepSeek Harness 的本地图像生成插件（优化版）：经本机 ComfyUI 调用 Qwen-Image 2.1 生成图像，全部流量限于 127.0.0.1。本版增强推理性能、NVIDIA/Intel/AMD 显卡适配与 Windows/Linux 兼容。原作者

Distributed image and video generation in ComfyUI for two NVIDIA DGX Sparks.

Serve your models fast on your own infra in 3 steps.

A fancy NVIDIA GPU monitor for the terminal. One Python file, no dependencies.

Automate ongoing market research with an AI planner, Jules agents, live conversations, and organized GitHub reports.

This project is inspired by CMP40HX-Unlock. If you find it useful, please consider giving a star to both the original author and this reposi

Local image generation and editing for NVIDIA DGX Spark, powered by Qwen-Image-2.1 and Gradio.

Terminal fleet monitor for GPUs, NPUs, and 13 other accelerator kinds, from utilization down to which pod is using them.

Local desktop tool for managing multiple SSH servers: GPU/process monitoring, SFTP file manager and high-speed server-to-server transfer (El

RTX 30 系 DLSS 帧生成管理器：按游戏部署 dlssg_for_sm86，一键恢复，自动检测内核级反作弊 | DLSS frame generation manager for RTX 30 series: per-game deployment of dlssg_fo

A local music generation and remix studio built on audio.cpp. Generate, repaint, split into stems and swap voices, with every take kept and

High-performance CUDA inference engine for Qwen models, written from scratch in C++/CUDA

Transactional cross-platform installer for DLSS5-Feeder, RenoDX, ReShade, and LumeniteFX

PantheonSim: GPU simulator for CUDA and ROCm. Run CUDA and HIP programs unmodified on a CPU, no GPU needed: simulated NVIDIA and AMD GPUs fo

Local-first AI workbench for OCR, transcription, translation, image generation, mind maps and composable workflows. Built for DGX Spark.

Durable local inference for Oh My Pi: NInfer + Qwen3.8 27B on one RTX 3090/4090/5090, with restart-resumable OpenAI Responses state.

Serve DeepSeek and Qwen, run an on-demand model library, and fine-tune with Unsloth QLoRA on 2x NVIDIA DGX Spark. TP=2 vLLM lanes behind one

A dashboard for a home cluster of NVIDIA GB10 (DGX Spark / ASUS GX10) inference servers: GPU and system health, plus live per-model metrics

Realtime TUI and web dashboard for multi-node NVIDIA DGX Spark clusters, with 200G ConnectX-7 RDMA fabric telemetry

Docker compose configs for serving Qwen3.8-27B on a DGX Spark (GB10): vLLM+MTP current, SGLang+DSPARK + FP8 rollback stacks

Serving Qwen3.8-27B-FP8 on a single DGX Spark (GB10): 7.88 to 58.5 tok/s single-stream from decode strategy alone, weights untouched. Specul

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

dgx-spark-qwen38 is an open-source installer that turns an NVIDIA DGX Spark, or any GB10 machine (ASUS, Dell, HP, Lenovo, MSI, GIGABYTE, Ace

A tiered-memory system design for workloads that don't fit in RAM: measure the working set, pin the hot tier, stream the cold tier from flas

See every GPU in your lab in the terminal, the browser or a report — who holds each GPU, which ones sit idle. One Python file, nothing to in

GLM-5.2 NVIDIA NIM Go 客户端 / OpenAI 兼容反向代理 — 自动化 hCaptcha 凭证池 + 流式推理，支持 Docker 部署。Go client & OpenAI-compatible proxy for NVIDIA Playground's

Headless Steam Big Picture in a rootless Podman container, streamed to a Steam Deck or any Moonlight client. Shares your game library, not y

Convert NVIDIA ARDY motion output (.npz) to BVH/FBX for Blender, Unity, Unreal

Fast, offline OCR for Node.js & C++. PP-OCRv6 with Core ML / WebGPU hardware acceleration — recognize text in images with confidence scores

Live per-node GPU, host, and vLLM stats for DGX Spark clusters, in your terminal

Catch local LLMs spilling into the CPU. One run shows GPU placement, prefill, decode, memory, power and a verdict.

Free NVIDIA DLI course: build, coordinate, and secure AI agents with NemoClaw and OpenShell. Browser-based, no GPU required. Available in En

MCP server for NVIDIA GPU metrics give AI agents real-time access to GPU utilization, memory, temperature, and power

Distill teacher chains-of-thought into a LoRA adapter via a strict boxed-answer format contract + two-phase Train→Nudge (silver-medal NVIDIA

Overclock or undervolt your NVIDIA RTX GPU from the command line — a no-background-process MSI Afterburner alternative for Windows.

Self-hosted vLLM made simple. Git-friendly YAML for models and profiles, full lifecycle from an intuitive CLI and TUI.

Offline-first Python AI agent that runs a tiny research business: quotes each job against its own costs, collects via Stripe, fulfils with N

Hands-on GPU/HPC infrastructure operations: K8s GPU scheduling, HAMi sharing, Slurm, observability & vLLM inference. Learn it free on a lapt

AI Auto Annotation & YOLO Training Pipeline, End-to-end object detection auto-labeling and YOLO training platform. VLM-powered annotation wi

The old advice to avoid Windows for local LLMs used to be right. It isn't anymore. The fixes for display-GPU desktop crashes and WSL memory

CPU-side action routing, context compression, and causal memory for AI agents — matches an LLM-everything agent at 58% fewer LLM calls and ~