Havuz

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

Can't find subtitles? Generate SRT/ASS subtitles from any video with local AI and auto-translate them — no internet, free. macOS & Windows.

Your coding agent starts every session with amnesia — memo fixes that, 100% on your machine. Persistent memory for Claude Code, Codex, Curso

Winning Qwen and DeepSeek setups for AMD Strix Halo: pinned Ansible recipes and matched benchmarks on a 128 GiB Ryzen AI Max+ 395 / Radeon 8

The best model under 200B at near 10 token/s exact batch throughput on a single laptop CPU, with an automatic 8 GB RAM path. Native C, no GP

QwenCleo-ASR is the currently State-Of-The-Art (SOTA) open-source model for Egyptian Arabic & code-switching speech recognition Built on Qwe

Interactive AI avatar live streaming with Qwen SSE, CosyVoice3 voice cloning and MuseTalk 1.5 lip-sync. Chrome chat collection, autonomous h

Companion code for the Manning book 'LLM Customization and Fine-Tuning.' Adapt open-weights LLMs end to end: prompting and RAG, LoRA/QLoRA,

A minimal yet ready-to-use AI agent harness, built on best practices for frontier models.

verl for a single consumer GPU. PPO, GRPO and on-policy distillation on NVIDIA GPUs.

To know what models don't say out loud.

Turn any document or a whole zip into an interactive knowledge graph, using a self-hosted Qwen3.6-35B-A3B-MTP on a single NVIDIA L4

Proxy API OpenAI-compatible que usa automação com Playwright para rotear requisições para modelos do Qwen com suporte a múltiplas contas, to

Fully local vulnerability research pipeline - 14B code-specialized LLM reviews every source file exhaustively.

Give any AI agent a full desktop — it sees the screen, clicks, types, and runs apps like a human. Automate anything with a UI: browsers, leg

Clone a voice from a few minutes of audio and generate speech locally: Qwen3-TTS fine-tuning pipeline with CLI and web UI

Fastest MoE/LLM inference runtime for consumer and edge Blackwell GPUs. SN74 on Gittensor.

Multi-Agent AI Desktop Client — 20 specialists auto-collaborate on your tasks. Visual Skill Store, PM orchestrator, streaming chat. Works wi

Self-hosted voice gateway in Go that speaks the OpenAI Realtime (GA) WebSocket protocol and runs a cascaded ASR → LLM → TTS pipeline behind

hummingbird is a lightweight, zero dependency runtime for massive open source Mixture of Experts (MoE) language models. It unifies SSD, RAM,

Local LLM-maintained personal wiki, built on Karpathy's LLM-Wiki pattern. Runs on Ollama + Qwen3 + QMD.

Airgapped closed-corpus QA loop: a self-hosted Qwen3.6 agent explores a .zip dataroom under a token budget with local tools

Inference engine for large language models on consumer GPUs, with deep optimization for older hardware

🚀 Multi-agent orchestration for Qwen Code CLI. 24 specialized AI agents and 82+ skills working as a professional dev department. Built for A

qwen2API çok dilli (i18n) fork'u • Multi-language (i18n) fork of qwen2API — WebUI artık Çince, İngilizce ve Türkçe / UI now in Chinese, Engl

The most powerful open-source TradingView AI agent. 50 tools, 15 models (Claude, GPT, Gemini, Qwen, DeepSeek, MiniMax), Pine Script v6, quan

The official Qwen Studio app only supports Windows and macOS. This project brings the same desktop experience to Linux.

Hemmingway-1: a 27B model that writes the way a person writes. Open weights, free for non-commercial use (CC BY-NC 4.0); commercial use by a

Local LLM inference engine written from scratch in Rust — hand-written AVX-512 assembly kernels, Metal & Vulkan compute shaders. Supports Qw

Free, open-source AI coding agent for terminal and browser. A Claude Code / Cursor alternative that runs offline with Ollama or with your ow

Local-first Model Context Protocol (MCP) memory layer for Codex CLI/Desktop, Claude Code, Gemini CLI, Qwen/DeepSeek/Ollama and agent workflo

Crow — a local AI agent, not a chat box: 30 tools, MCP, memory, skills, goals, subagents, vision, and image generation and editing with Qwen

Ultra cost-effective terminal AI coding assistant with excellent token cache hit rate. Built in ~10K lines of Go, it uses DeepSeek by defaul

qMLX: Custom inference engine for Qwen 3.5 122B on Apple Silicon, extending MLX with hybrid attention support, SSD-backed KV cache, and RYS

Pixel-Pets: A local, open-source virtual pet ecosystem on M5Stack. 100% AI-authored C++ (Claude) by a 10yo maker. Includes Muffin (LLM), Vis

阿亮开源的 Skills 合集：短剧成片 / 电商详情图 / 儿童故事 / 有声绘本 / 配音 / 多人播客。一句话触发，AI 走完整 AIGC 流程。

让 Codex 白嫖 NVIDIA 免费顶级模型，一键接入 | Transparent proxy between Codex and NVIDIA NIM — use NVIDIA's free hosted top-tier models with Codex

[Archived] AAOSP, predecessor to JibarOS. Early experiment in OS-level on-device AI (MCP + local LLM on Android 15). See JibarOS for the act

🦄 Forge - a local-first, multi-agent software-engineering runtime that runs Claude Code / Codex-style agentic workflows entirely on your own

Run a 177B model (Qwen3.8-Flash-Next) on an 8 GB laptop GPU. Measured: 6.6 GiB VRAM, 47.8 GiB RAM, 34-35 tok/s. The n-gram table stays on di

A NPC AI Framework , based on character personalities, can be used for game NPC (extendable for Godot / Unity)

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;

AI model that animates any SVG into Lottie with 3-4x fewer tokens than OmniLottie (CVPR 2026). Upload SVG → get animated .lottie. 3B model,

BrighTO LLM Router: free open-source, ultra-fast Rust LLM gateway/router and LiteLLM/Bifrost alternative with OpenAI + Anthropic APIs, Claud

3D explorable atlas of every neuron in a transformer LLM. Click neurons, trace residual highways, compare cross-architecture conservation. L

网页版登录态 → OpenAI 兼容本地反代：DeepSeek / Qwen / Doubao 免费额度驱动 ZCode。Web login → OpenAI-compatible local API.

What I run on my inference server: Qwen3.8-Flash-Next on two RTX 3090s, and how I got here

👻 MoEWatch: A lightweight diagnostic and audit library for Mixture-of-Experts (MoE) models in HuggingFace Transformers — the pytest for MoE

Instinct: single-forward-pass decision models (choice / true-false / score) returning a probability per candidate

Full-pipeline ComfyUI node pack tailored for Krea 2 Turbo: NAG negative guidance, Variance diversity, custom schedulers, samplers & aestheti