Trending AI projects

A Model Context Protocol server that turns Figma design systems into an API, enabling AI assistants to create designs, sync tokens, and audit design-code parity.

OpenAgentsControl (OAC) is an open-source layer on top of OpenCode that makes AI coding agents learn your project's patterns and conventions, then propose plans you approve before they write code, cutting refactoring and token waste.

Future AGI is an open-source platform that unifies simulation, evaluation, guardrails, tracing, and optimization for AI agents into a single feedback loop, aiming to make production agents reliable and self-improving.

A system to run, manage, and scale AI workloads across any AI infrastructure, providing a unified interface to avoid vendor lock-in and optimize GPU utilization.

An enterprise-ready orchestration framework and SDK for building, orchestrating, and deploying AI agents and multi-agent systems across multiple LLM providers.

SparkDiffusion is an open-source framework for accelerating video generation with Diffusion Transformer (DiT) models, targeting Wan 2.1 and Wan 2.2. It combines sparse low-rank attention, few-step distillation, and custom operators to speed up inference by over 200x while preserving quality.

NOFX is an open‑source, self‑hosted trading terminal that runs crypto‑perpetual bots powered by large language models. A Go runtime enforces strict risk limits before any model‑suggested order reaches one of nine supported exchanges. The web UI shows live positions, model reasoning, and lets users edit prompts, indicators, and leverage. Multiple bots can be compared on a public leaderboard. Install via a one‑line script, Docker, or manual build; supports many LLM providers (OpenAI, Claude, Gemini, etc.) and a proprietary pay‑as‑you‑go model service (Claw402).

A Spring-friendly API and abstraction layer for Java developers to integrate various AI models and vector stores into enterprise applications with portability and modularity.

A Chrome extension that turns YouTube videos into learning resources by providing AI-generated summaries, bilingual transcripts, and timestamped notes using the user's own API keys.

A comprehensive open-source guide and code repository for understanding Reinforcement Learning from Human Feedback (RLHF) and the post-training of language models.

A visual workspace for collaborating with AI coding agents, featuring WYSIWYG editors, parallel session management via git worktrees, and integrated task tracking.

A free, MIT-licensed question bank and study kit for AI job interviews: 28 topic areas, 720 illustrated questions, 11 course-style PDF decks, and role-specific reading routes — with a strict rule that answers emphasize engineering trade-offs and verifiable, real data.

An automated software development factory that uses AI agents to manage the full lifecycle of issues, from planning and implementation to rigorous verification and merging.

A typeface and encoding system that protects written work by serving decoy text to AI scrapers while rendering the original words for human readers.

A DSH plugin that drives DeepSeek agents using your authenticated web session instead of an API key, allowing free access to advanced model features.

A C++23/GGML implementation of NVIDIA's MotionBricks that enables high-performance skeletal animation and motion inference on CPUs and Vulkan-compatible hardware.

A desktop GUI for the pi coding agent that enables parallel task execution in git worktrees and integrated code review and terminal access.

ollaya is a Rust‑based tool that serves non‑generative decision models locally, offering an Ollama‑like CLI, TypeSafe‑compatible HTTP API, and support for ONNX and GGUF back‑ends. It ships tiny ONNX graphs, pulls weights from Hugging Face, and provides calibrated, millisecond‑level predictions for structured questions, targeting AI agents and decision‑making workloads.

OpenAver is a local, desktop-based metadata scraper and media manager for JAV collections that organizes files, fetches metadata, and provides a visual browsing interface with AI-ready API capabilities.

An agent harness for Codex that integrates the OmO engine to provide project memory, strategic planning, and verified execution for complex codebases.

GenieX is an on-device inference runtime for Qualcomm Snapdragon devices that allows developers to run LLMs and VLMs locally using a unified SDK across CLI, Python, Android, and server interfaces.

A free 11-module course teaching social media automation through the ViralWave Studio AI tool — covering bulk content generation, AI video creation, multi-platform publishing, analytics, and optional API workflows. Claims dramatic time savings (90%) and engagement gains, but the README is promotional and results are self-reported.

An open-source orchestration framework for real-time digital-human conversations, integrating LLMs, TTS, STT, and video rendering via WebRTC.

This project provides a Model Context Protocol (MCP) server that connects AI assistants like Codex and Claude to a local Adobe Premiere Pro installation. It exposes 283 editing tools, context resources, and prompts, allowing models to inspect, edit, and export projects through a CEP panel bridge.

Vexa is an Apache‑2.0 open‑source platform that runs bots which join Google Meet, Teams, Zoom (and soon Jitsi) calls, capture speaker‑attributed live transcripts, store them as Markdown in a Git repo, and let sandboxed AI agents read/write that repo. It offers a unified API and a web Terminal UI, can be deployed via Docker or Kubernetes, and is designed for self‑hosted, air‑gap‑ready use cases.

RuoYi AI is an open‑source, enterprise‑grade AI assistant platform. It combines multi‑model LLM access, local RAG with Milvus/Weaviate/Qdrant, a Langchain4j‑based multi‑agent framework, and a drag‑and‑drop visual workflow designer. Built on Spring Boot 3 (backend) and Vue 3 (frontend), it ships with Docker compose files for one‑click deployment of the backend, admin UI, user UI, and supporting services (MySQL, Redis, vector DBs, MinIO). The project provides separate repos for backend, admin panel, user frontend, and demo apps, and is licensed under MIT.

A Python library for financial econometrics that provides tools for modeling volatility, testing for unit roots and cointegration, and performing statistical bootstrapping.

A completely offline voice-to-text input tool for Windows that enables low-latency, push-to-talk typing and file transcription using local ASR models.

BettaFish (微舆) is an open‑source, Python‑based multi‑agent platform that automatically crawls, analyses and reports on public opinion from dozens of social‑media sources. Users ask a question in a chat‑like UI; three specialized agents (Query, Media, Insight) gather data, a ForumEngine lets them debate under an LLM moderator, and a ReportEngine assembles an interactive HTML (or PDF/Markdown) report. The system includes a 24/7 crawler (MindSpider), multimodal video/image processing, fine‑tuned sentiment models, and supports private‑data integration. It ships with Docker compose, a Flask front‑end, Streamlit demos, and is extensible to other domains.

A collaborative speedrun to find the fastest way to train a language model to a specific loss target on 8xH100 GPUs, reducing training time from 45 minutes to under 3 minutes.

dsh‑memory‑evolve is a DSH plugin that gives the AI persistent cross‑session memory, todo/project logs, Git‑based memory sync, multi‑agent orchestration (internal sub‑sessions and external AI providers), a searchable infinite canvas for files, optional reviewer sessions, and mobile‑friendly UI with notifications. It lets the AI act as a long‑term partner and team coordinator while keeping the user in control of all writes.

A curated list of open-weight uncensored LLMs for authorized red team operations, penetration testing, and security research, with detailed specs, training data, and deployment guidance.

A Python library to fetch YouTube video transcripts (including auto-generated ones), translate them, and format them into various output formats. It works without a browser and supports proxies to avoid IP bans.

A comprehensive automation platform for bulk-registering and managing AI service accounts, featuring lifecycle monitoring, proxy rotation, and integration with API gateways.

An MCP server that enables AI assistants to interact with Obsidian vaults by reading, searching, and editing notes via the Local REST API plugin.

Kun is a local‑first AI desktop workspace that merges code editing, document management, and chat‑style agents into one Electron app. All data stays on your computer; you connect any OpenAI‑compatible, Anthropic‑compatible, or local model (e.g., Ollama, DeepSeek). Features include project‑wide code assistance, a research‑paper workspace, extensible plugins, and a terminal UI. It’s aimed at developers and researchers who want AI‑augmented productivity without cloud‑side data storage, and it’s released under a non‑commercial PolyForm license.

dlt is an open-source Python library that automates data loading by extracting data from various sources and moving it into structured, typed destinations. It simplifies pipeline creation by handling schema inference, type mapping, and database-specific requirements automatically.

Source content for Modal's GPU glossary, with dual licensing (CC BY 4.0 for glossary text, MIT for the rest) and a community Chinese translation.

A curated gallery of Nano-banana/Nano-banana-pro generated images with their prompts, organized into 52 Pro and 110 standard examples, plus a 150k-image dataset for identity-consistent editing.

A curated collection of high‑quality MuJoCo robot models (humanoids, quadrupeds, arms, drones, etc.) packaged as a Python library for easy loading and simulation.

A local-first, browser-based AI video editor featuring a multi-track timeline and WebGPU-powered AI for music, voiceovers, captions, and video restoration.

HiRoute is a local model router that reduces costs for AI coding agents by sending routine tasks to cheaper models and keeping premium models for difficult work. It runs as a desktop app or headless CLI, supports difficulty-based routing, custom branches, and tracks usage and cost per session.

Obsidian Copilot is an open‑source Obsidian plugin that adds AI agents (opencode, Claude Code, or Codex) for multi‑step research, writing, and knowledge‑work. Agents can read/write notes, run tools, and keep context across turns. You can use hosted models, bring‑your‑own‑API keys, or local models; free use works with your own provider, while paid plans unlock hosted services and multi‑agent features.

Marin is an open‑source platform for end‑to‑end development of foundation language models, offering declarative experiment pipelines, scaling‑law tools (Delphi), mixture‑of‑experts support, and extensive documentation, backed by Stanford CRFM, Open Athena and major research sponsors.

Airbyte is an open-source data movement platform that syncs data from hundreds of sources to warehouses or AI agents, providing an SDK to turn these connectors into LLM-ready tools.

A daily-updated, AI-verified leaderboard of the top 100 AI-agent skills, merging install counts from multiple registries and social buzz into nine rankings, with raw CSV data published openly under a permissive license.

A set of Codex Skills that transforms raw business materials into professional, consulting-style PPT slides using a synthesis-first workflow and a separate image-to-editable conversion tool.

VGGT‑Ω is a research‑grade neural network that jointly predicts camera pose, depth maps, and (optionally) text‑aligned embeddings from RGB images. The repo provides pretrained 1‑B‑parameter checkpoints (512 px, 416 px reproduction, 256 px text‑aligned), training/curation code, data‑preparation guides, a Gradio demo, and detailed GPU‑memory benchmarks. It targets 3‑D reconstruction, AR/VR, robotics, and other geometry‑aware vision tasks.