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
illuin-tech
A vision language model approach (and colpali-engine library) that retrieves documents straight from page images, replacing brittle OCR pipelines with ColBERT-style late interaction (MIT).
langchain-ai
LangChain's low-level orchestration framework for building durable, stateful, long-running agents with human-in-the-loop control and persistent memory (MIT).
ZiYang-xie
Open-source project that generates full, 360-degree explorable 3D scenes in seconds from a text prompt or single image, for games, simulation, and VR (Apache-2.0).
suno-ai
Suno's fully generative text-to-audio model — speech, music, and sound effects from one transformer, with nonverbal cues like [laughs] and 100+ voice presets (MIT).
RVC-Project
The de facto open-source voice-conversion framework — trains a usable voice clone from 10 minutes of audio via top-1 feature retrieval, with a ~170ms real-time GUI (MIT).
argmaxinc
Argmax's Swift SDK bringing on-device Whisper speech-to-text (plus TTS and speaker diarization) to Apple Silicon — CoreML-accelerated, no audio leaves the device (MIT).
deepseek-ai
DeepSeek's open full-stack codebase for training and evaluating speculative-decoding draft models — data prep, DSpark/Eagle3 training, nine eval benchmarks, and released checkpoints (MIT).
allenai
Ai2's open toolkit that converts PDFs and scanned documents into clean Markdown with a 7B vision-language model — equations, tables, handwriting, under $200 per million pages (Apache 2.0).
diegosouzapw
Open-source AI gateway connecting Claude Code, Cursor, Cline & Copilot to 237 providers (90+ free) through one endpoint, with auto-fallback and 15–95% token compression (MIT).
mudler
Open-source, self-hosted AI engine with a drop-in OpenAI-compatible API that runs LLMs, vision, voice, image, and video models on any hardware — no GPU required (MIT).
wgsxm
Open-source structured 3D generative model that produces part-separated meshes from a single RGB image using a compositional latent diffusion transformer (NeurIPS 2025, MIT).
Kortix AI
Source-available AI agent platform (the engine behind Kortix) that models a company as a versioned git repository of owned, self-hostable agents, skills, and memory.