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.
760 projects
iOfficeAI
Apache-2.0 Office suite built for AI agents — read, edit, and automate Word, Excel, and PowerPoint from a single dependency-free binary, with a built-in HTML/PNG rendering engine that lets agents visually verify output.
stablyai
Open-source Agentic Development Environment (ADE) that runs Codex, Claude Code, OpenCode, and Pi in parallel — each in its own git worktree — with a mobile companion, on macOS/Windows/Linux (MIT).
browser-use
MIT-licensed open-source pipeline that edits videos through a coding agent — drop in raw footage, chat with Claude Code, and get a finished cut with filler-word removal, color grading, subtitles, and animation overlays.
baaivision
BAAI's open multimodal suite that tokenizes image, text, and video into one discrete space and trains a single transformer with pure next-token prediction — generation and perception, no diffusion or CLIP (Apache-2.0).
pyannote
The de facto open-source speaker diarization toolkit — PyTorch-native "who spoke when" pipelines with state-of-the-art pretrained models on Hugging Face, fine-tunable on your own data (MIT).
facebookresearch
Meta FAIR's library of efficient, reusable, GPU-accelerated components for 3D deep learning — batched mesh/point-cloud data structures, differentiable operators, and a differentiable renderer for PyTorch (BSD).
facebookresearch
Meta AI's third-generation self-supervised vision foundation model — frozen backbones from 21M to 6.7B params that deliver high-quality dense features and beat specialized SOTA without fine-tuning.
kyutai-labs
Kyutai's full-duplex speech-text foundation model for real-time voice dialogue (~200ms latency), built on the streaming Mimi neural audio codec — Apache-2.0 code, CC-BY-4.0 weights.
SparkAudio
An efficient LLM-based text-to-speech model built entirely on Qwen2.5 with BiCodec single-stream tokens — zero-shot voice cloning and controllable voice creation in Chinese and English (Apache-2.0).
AI4Finance-Foundation
AI4Finance's open-source project for efficiently fine-tuning LLMs on financial data — powering low-cost sentiment analysis, forecasting, and robo-advisory (MIT).
milvus-io
High-performance, cloud-native open-source vector database for billion-scale ANN search — the storage layer behind RAG, semantic search, and recommendation (Apache-2.0).
xorbitsai
Open-source framework to serve LLMs, speech, embedding, and multimodal models behind one OpenAI-compatible API, backed by engines like vLLM and llama.cpp (Apache-2.0).