Open Source
Explore the latest AI open-source projects from GitHub and HuggingFace.
Explore the latest AI open-source projects from GitHub and HuggingFace.
Explore the latest AI open-source projects from GitHub and HuggingFace. Discover projects across categories like LLM, Vision, Audio, and more.
886 projects
lokutor-ai
Open-vocabulary English speech recognition running on a $5 ESP32-S3, with 3.7% WER on LibriSpeech test-clean
SupraLabs
English single-speaker Glow-TTS + HiFi-GAN trained on LJSpeech, MIT-licensed; UTMOS 3.59 vs 3.05 for its predecessor Flare-TTS-v1.5.
timoncool
Native Windows app that makes full songs with vocals and an editable sheet-music score on your own GPU, using the YuE2 model.
Altworld
Altworld's 27B writing-focused fine-tune of Qwen3.8-27B with a 262K context, built to return the message itself; CC BY-NC 4.0 (non-commercial).
PSRben
Visual backbone where memory and its update rule co-evolve within an image; T/S/B models for ImageNet, COCO and ADE20K
Tencent
Tencent's Go knowledge framework combining RAG Q&A, a ReAct agent with sandboxed skills and MCP, and an auto-generated wiki; self-hostable with Docker.
TencentARC
Geometry-native latent space that decodes RGB and 3D geometry together for camera-controlled video and point-cloud generation.
Niko1221
C++ engine that runs the 125B MoE Qwen3.8-Flash-Next on one consumer NVIDIA GPU plus RAM, with a one-click Windows/Linux install.
zai-org
Z.ai's coding agent harness: Electron desktop app, web client and terminal UI with a plugin, MCP and hook system, source-built with pnpm.
Edge0
Streaming Chinese/English ASR with 80-160 ms clock, 240-560 ms delay and rolling-KV 24/7 transcription via an adapted vLLM build
nimaone
Offline Persian text-to-speech with voice cloning on CPU, with a torch-free ONNX Runtime path and FastAPI web demo. Weights are CC-BY-NC-4.0.
Liuziyu77
Multimodal System One decision model: text, images and video in, probabilities over candidates out, on a Qwen3.5 0.8B/2B backbone.