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
FluidInference
Swift SDK for local ASR, TTS, VAD, and speaker diarization on Apple devices, using open-source CoreML models on the Neural Engine.
XiaomiMiMo
Xiaomi's flagship 1.02T-param MoE (42B active) with native text/image/video/audio input and a 1M-token context, tuned via large-scale RL.
leejet
Pure C/C++ engine (ggml-based, like llama.cpp) running Stable Diffusion, FLUX, Qwen Image, Z-Image, and Wan video models locally.
surya-koritala
Rust codec that holds AI model weights and KV cache compressed in GPU memory, decoding bit for bit; also a general LZ4/zstd alternative.
strands-agents
Open-source Python/TypeScript SDK for AI agents, with both a low-level agent loop and a batteries-included harness with benchmarked defaults.
TypeLLM
Type-safe generation library for existing LLMs served on SGLang, adding schema-guaranteed outputs; inspired by TypeSafe AI's Jev.
apple-aiml-research
9B Apple VLM that scans compressed page images and re-expands only relevant pages via a learned tool call, keeping near full-text accuracy.
impossibleG
Local-first Windows dictation and meeting app running Vosk or Whisper on-device, MIT-licensed, with local history and optional Ollama formatting.
harry7557558
Self-contained, cross-vendor 3D Gaussian Splatting app: trains photos or video into a textured mesh, with built-in SfM and AI masking.
filliptm
ComfyUI nodes for YuE2-3B music generation with an editable piano-roll ABC score editor and AR/joint AR-NAR LoRA training support.
oddadmix
A 7.48M-param, 28.7 MB distillation of hexgrad Kokoro-82M for offline English TTS, 4.1x faster than the teacher and ~45x realtime on 4 CPU threads.
inclusionAI
Two 6B open-source (MIT) models for text-to-image design generation and RGBA layer decomposition of design images, from inclusionAI.