Trending AI projects

NeMo Gym is NVIDIA’s open‑source library for scalable evaluation and training of AI agents in stateful environments (code execution, tool calling, safety, etc.). It provides modular interfaces, a hub of ready‑made benchmarks, support for thousands of concurrent runs, and integration with any RL framework. Install via `uv`, configure a model (e.g., OpenAI), start the environment servers, and run the `gym eval` CLI to collect rollouts and metrics.

Ultra Flash is a framework for real-time high-resolution streaming video generation, enabling 1K and 2K video output on a single GPU through a cascaded architecture of sparse super-resolution and dynamic cache optimization.

A tool that transforms text or images into interactive 3D holographic and lenticular trading cards with a Blender-powered pipeline and a Three.js web viewer.

A governed shared-memory layer for AI agent fleets that allows multiple agents to store, recall, and compound knowledge across teams and organizations.

A runtime engine that enables running Mixture-of-Experts models larger than a device's RAM by streaming experts from flash storage, allowing massive LLMs to run on smartphones and CPUs.

Memoh is an open-source multi-agent platform that provides each AI agent with its own persistent cloud-based computer, enabling them to run 24/7 with dedicated access to files, browsers, and long-term memory.

AI Story is a Docker‑based web platform that turns a short text prompt into a complete short video. It automatically rewrites the prompt into a script, creates a storyboard, generates images via multiple AI image services, plans camera motions, and renders the final video, all orchestrated with Django, Celery, Redis and a Vue front‑end.

A portable agent skill that enables AI agents to generate interactive, self-assembling 3D architectural miniatures using Three.js and procedural materials.

pyannote.audio is an open-source Python toolkit for speaker diarization, offering state-of-the-art pretrained models and pipelines that can be further fine-tuned for improved performance. It also provides a premium cloud service via pyannoteAI.

A toolkit of AI agent skills that applies professional filmmaking and dramaturgy principles to generate high-precision prompts for AI video and image models.

Claude-Skills is a library of 385 ready-made skills that turn AI assistants into specialists for tasks like board preparation, product planning, compliance audits, code review, and marketing. It bundles Python tools for accurate computation and works across 11 AI assistants, with an ask-before-write approach to avoid guesswork.

GithubStarsManager is an Electron‑based desktop app that syncs your GitHub stars, enriches each repo with AI‑generated summaries, tags and categories, and lets you search them semantically via Cloudflare Vectorize. It also offers a repository Q&A assistant, release‑tracking timeline, fork and Gist management, a discovery feed, and an optional self‑hosted backend for cross‑device sync. Supports multiple LLM providers, local‑first data storage, and a Model Context Protocol server for AI agents.

An open-source AI novel writing studio that helps authors plan worlds, characters, and plot, and execute chapter-by-chapter production with a specialized AI Director.

An AI runtime that turns your own PC into a local OpenAI-compatible API server, supporting LLMs, speech, image and audio generation across CPU, GPU and NPU backends.

A modular agent framework for automating cyber-security workflows, including security scanning, code auditing, and binary analysis.

SEO-pro is a ChatGPT plugin that turns page URLs, drafts, or Search Console exports into sourced, prioritized SEO fixes with verification steps. It works directly inside ChatGPT, requires no API key or external SEO account, and uses only the evidence you provide.

Awesome RSI is a community-curated list of research on AI systems that improve how they improve themselves, organized into a detailed taxonomy, with careful definitions of what counts as genuine recursive self-improvement.

A public collection of over 100 ready‑to‑import AI “persona skills” (prompt templates) for agents, covering writers, celebrities, historic texts, social‑media influencers, and multi‑agent teams. Compatible with OpenClaw/StepClaw and the nuwa‑skill format, it lets developers instantly give LLMs a distinct voice or run collaborative agent crews.

Mnemosyne is an open‑source, SQLite‑backed memory layer for LLM agents. It offers a local‑first, zero‑cloud solution with a Python SDK, a built‑in MCP server, and optional encrypted sync. The system stores facts, embeddings, and temporal triples using a three‑tier BEAM architecture (working, episodic, triple‑store) and performs hybrid vector‑keyword search inside SQLite. It integrates with many agent platforms (Claude Code, Cursor, OpenWebUI, Hermes, etc.) and runs with a single `pip install` and minimal resources.

A browser automation framework for AI agents that bypasses anti-bot protections and provides token-efficient web interaction via indexed text formats.

A GPU-native library providing foundation models and tensor subclasses for efficient processing and prediction on tabular and relational structured data.

A collection of standardized skill packages that provide AI agents with specialized knowledge and best practices for developing with Supabase and Postgres.

An AI-powered card note-taking application that uses RAG to enable natural language retrieval of self-hosted notes.

An open-source collaborative OS for AI agents that provides a unified workspace where multiple agents can share threads, files, and a browser to work together.

Cache-to-Cache (C2C) is a framework that enables LLMs to communicate directly by projecting and fusing their internal KV-Caches, bypassing the need for text generation to improve speed and accuracy.

An agent-native operating system that replaces traditional menus and folders with a conversational AI interface powered by Hermes Agent to manage files, apps, and system tasks.

An open-source agentic AI data assistant that connects to databases and files to autonomously write SQL and code for data analysis and reporting.

PyMuPDF is a high-performance Python library for PDF and other document processing, offering fast text extraction, layout analysis, conversion, rendering, and manipulation. It provides LLM-ready outputs like Markdown and JSON, runs locally, and supports OCR, tables, forms, and redaction.

A research-oriented framework for developing and testing autonomous agents within a simulated MMO environment, designed for studying goal-directed program synthesis and multi-agent systems.

A 1-click launcher for open-source projects that provides a user-friendly UI to automate the installation and execution of terminal-based software.

A specialized decision model for agentic AI that performs fast, reliable classification and scoring tasks without the overhead of full text generation.

Model Context Protocol (MCP) is an open standard that provides a universal way for AI models to connect to data sources and tools, reducing the need for custom integrations.

An AI-native API gateway based on Istio and Envoy that provides unified LLM API management, MCP server hosting, and production-grade traffic control for AI workloads.

A terminal application that manages multiple AI coding agents in isolated git workspaces, allowing developers to run several tasks simultaneously without conflicts.

The MCP TypeScript SDK provides TypeScript libraries to build Model Context Protocol (MCP) servers and clients, enabling standardized, type‑safe delivery of tools, resources, and prompts to LLMs. It runs on Node, Bun, and Deno, offers middleware for popular web frameworks, and includes docs, examples, and migration guides.

AutoSubs is a cross‑platform desktop app that generates offline subtitles from audio/video, offering speaker labeling, many local transcription models, and direct integration with DaVinci Resolve and Adobe editing tools.

NVIDIA NeMo Guardrails is an open-source toolkit that adds programmable safety and control layers to LLM-based conversational apps. It uses configurable rails and a Python-like dialogue language (Colang) to prevent unwanted topics, detect jailbreaks and injections, fact-check outputs, and manage tool calls, all while integrating with popular LLMs and LangChain.

A comprehensive manga translation tool that automates text detection, OCR, translation, and typesetting using AI models and a visual editor.

A load balancer and proxy for ChatGPT accounts that enables request distribution, usage tracking, and API key management via an OpenAI-compatible interface.

An enterprise-grade agentic workflow development platform that integrates AI orchestration, RPA automation, and model management for production-ready AI agents.

openTPU is an open‑source AI accelerator built on a Xilinx Kintex‑7 FPGA. The repo includes hardware RTL, a custom 32‑bit ISA, a Python simulator, a kernel language/compiler, and host tools for chatting, serving, monitoring and profiling. It runs real LLMs (230 M‑35 B parameters) on a PCIe card, achieving up to ~86 tokens/s with 4‑bit quantisation and DRAM utilisation near the DDR3 peak. The project is both a functional inference platform and a learning resource for hardware‑aware AI research.

A visual workspace and CLI agent manager that organizes AI conversations and project assets into a structured desktop and mobile interface.

A local, voice-controlled assistant for Windows that automates PC tasks, smart home devices, and web services using a mix of local and cloud-based LLMs.

A Chinese translation of the book *Agentic Design Patterns*, covering AI agent design patterns. All 32 files are translated and under review. Contributions are welcome via GitHub pull requests.

An open-source library for ingesting and pre-processing unstructured documents like PDFs and Word docs into structured formats for LLM workflows.

A Rust-based MCP server and CLI that allows AI agents to control a Linux desktop by reading accessibility trees, taking screenshots, and simulating user input across multiple window managers.

GR00T N1.7 is an open vision-language-action model that lets robots perform manipulation tasks from language and camera input. It combines a vision-language backbone with a diffusion action decoder, supports cross-embodiment generalization, and can be fine-tuned and deployed on edge devices.

A private, automatic work journal for Mac that uses AI to turn screen activity into a chronological timeline and searchable work history.