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
OpenMOSS
A 0.9B Apache-2.0 model that jointly transcribes and diarizes long multi-speaker audio in one pass, cutting speaker-attribution error roughly 10x versus commercial ASR on AISHELL-4.
MisoLabsAI
An 8.2B-parameter open-weights TTS model using an RVQ Transformer with a Llama-8B backbone and Mimi codec, built for dialogue-context-aware English speech and zero-shot voice cloning.
0xShug0
A pure C++ ggml inference engine that runs 35 audio model families — TTS, ASR, VAD, diarization, source separation, and music generation — with no Python dependency and CUDA speedups of up to 8x.
CVHub520
A GPL-3.0 desktop annotation tool with a built-in AI engine, using SAM 3, Grounding DINO, YOLO, and PaddleOCR to auto-label images and video across detection, segmentation, OCR, VQA, and tracking.
headroomlabs-ai
An Apache-2.0 local-first context compression layer for AI agents, cutting 60-95% of tokens on JSON and 15-20% on coding agents via content-aware, reversible compressors.
alibaba
Alibaba's production-hardened AI code review CLI, combining deterministic pipelines with an LLM agent for precise line-level comments at roughly one-ninth the tokens of general-purpose agents.
ruvnet
An MIT-licensed, Rust-based platform that turns commodity WiFi Channel State Information from cheap ESP32 boards into contactless sensing — presence, vital signs, pose, and activity — with Home Assistant and Matter integration.
dottxt-ai
A mature Apache-2.0 Python library for structured LLM generation that constrains output at the token level to match Pydantic models, JSON, regex, choices, or grammars, working across local models, servers, and hosted APIs.
microsoft
A Microsoft open-source optimizer that trains reusable natural-language skills for frozen LLM agents through trajectory-driven, validation-gated edits, producing compact best_skill.md artifacts with no added inference cost.
RunanywhereAI
A source-available toolkit for running LLMs, vision, speech-to-text, and text-to-speech models fully on-device across iOS, Android, Flutter, React Native, and Web, with Qualcomm Hexagon NPU acceleration.
1jehuang
A next-generation, Rust-built coding agent harness focused on multi-session workflows, low resource use, and semantic agent memory, with cross-platform installs and provider-agnostic model support.
Tencent-Hunyuan
Tencent Hunyuan's open text-to-3D human-motion model: a Diffusion Transformer with Flow Matching that generates skeleton-based character animations from text prompts, scaled to the billion-parameter level with 1.0B and 0.46B variants.