Trending AI projects

A curated directory of AI tools, frameworks, and datasets designed to accelerate scientific research, covering everything from literature management and document parsing to autonomous research agents.

A single-person maintained directory of free LLM APIs, curating both providers and recommended models. Only OpenAI-format endpoints are listed, with per-provider rate limits and model names. Selection prioritizes long-term stability, rate-limit style, source diversity, and model quality. Community submissions are welcome, and the visual table is LLM-generated.

dsh‑commandcode‑provider is an unofficial DeepSeek Harness plugin that adds Command Code as an LLM provider. It offers UI integration, multi‑account rotation, plan‑aware model picking, vision support, cost estimates, quota display, web‑search, and optional zero‑data‑retention, working with dsh 0.2.x.

An open-source AI marketing plugin providing a standardized 12-part strategy flow and 163 skills to help agencies and in-house teams manage large brand portfolios consistently.

LLM4AD Next is a Python library that uses a conversational LLM to turn a natural‑language problem description into a complete, runnable evolutionary‑algorithm search pipeline. It ships with multiple GA‑style search methods, modular “algorithm‑design skills”, an advisor that can suggest code blocks to evolve, and an online demo. Install via `uv sync` (Python 3.12+), configure an LLM provider, then run `llm4ad chat` to generate a project and `llm4ad run <config>` to evolve the best algorithm.

Firefox DevTools MCP is an open‑source Model Context Protocol server that lets AI agents control a local Firefox browser—including its developer tools—via WebDriver BiDi. It enables agentic browsing and AI‑assisted web development, exposing tools like navigation, form filling, network inspection, console access, and JavaScript evaluation. The server runs locally (via `npx @mozilla/firefox-devtools-mcp@latest`), works with Claude Code, Codex, Cursor, etc., and offers security‑focused presets to limit capabilities.

A curated list of ChatGPT-related GitHub repositories, with a search tool and integrations for AI coding assistants.

FastEmbed is a lightweight, fast Python library for generating embeddings (numerical representations of meaning) from text, images, and documents. It uses ONNX Runtime instead of PyTorch, runs on CPUs without a GPU, and supports dense, sparse, late-interaction, image, and multimodal embedding models, plus rerankers. It integrates with the Qdrant vector database and is designed for serverless environments.

SpecForge is a unified training framework for speculative decoding models that integrates directly with the SGLang serving engine to accelerate LLM inference.

A production-ready framework and codebase for building end-to-end LLM systems, covering training, RAG, and deployment on AWS.

A hybrid edge-cloud AI engine for mobile devices and wearables that optimizes inference for text, speech, and vision models using custom quantization and hardware-specific kernels.

A reference architecture for building GPU-accelerated AI agents that perform real-time video analytics, natural language search, and automated summarization using vision-language models.

An efficient closed-loop harness for embodied AI that evolves code-based recovery skills and runtime critics to improve robot task success rates without retraining base policies.

An end-to-end traffic analysis suite that creates 3D digital twins from CCTV footage and satellite maps using computer vision and a custom calibration pipeline.

JARVIS V4 is a desktop voice assistant for macOS and Windows that enables users to perform tasks using Google Gemini's AI capabilities via a personal API key.

An offline, privacy-focused Android tool that provides speech-to-text transcription and live captioning using local machine learning models.

A collection of standardized, modular skills and documentation tools designed to help AI coding agents perform Kotlin-specific development tasks more effectively.

A lightweight operations toolkit for homelabs that tracks server state changes, verifies backups, and provides structured diagnostics for humans or AI agents.

A curated collection of academic papers on humanoid robot learning, organized by task and highlighting research with real-world experiments and open-source code.

A multimodal MoE model that serves as an architectural preview for Qwen4, offering high efficiency in coding and office tasks with significantly reduced training and inference costs.

An open-source vulnerability scanner for LLMs and agent workflows that protects AI systems from jailbreaks, fuzzing, and multimodal attacks.

An AI-powered relationship analysis tool that processes WeChat and QQ chat logs to quantify romantic dynamics using psychological frameworks.

MOSS‑VL is an open‑weight, 11 B parameter video‑language model series that supports real‑time interaction on live video streams (MOSS‑VL‑Realtime) as well as offline high‑quality video QA (MOSS‑VL‑Instruct) and a base checkpoint for further training. It uses a cross‑attention architecture with timestamps and 3‑D rotary embeddings, offers interruptible streaming, proactive silence, and dynamic correction, and provides quantized checkpoints that run on a single 24 GB GPU. The repo includes ready‑to‑run inference scripts, a FastAPI/WebSocket server, SGLang‑based serving back‑ends, fine‑tuning scripts (full‑parameter and LoRA), and extensive benchmarking results.

A multi-platform package manager and inference interface for Stable Diffusion that provides one-click installation and shared model management across multiple Web UI packages.

An open-source AI voice input tool for macOS, Windows, and Linux that provides app-aware dictation, voice Q&A, and text rewriting.

Speculators is a Python library that trains small “draft” models for speculative decoding, enabling faster, loss‑less LLM inference when deployed with vLLM.

A library of plugins that enables AI coding agents to perform specialized tasks across the Adobe ecosystem, including analytics reporting, AEM development, and App Builder project management.

TinyTroupe is an experimental Python library that simulates people with specific personalities, interests, and goals using LLMs, enabling realistic multiagent interactions in simulated environments to gain insights for business scenarios like ad evaluation, product feedback, and synthetic data generation.

GMR is a general motion retargeting library that converts human motion data (from SMPL-X, BVH, FBX, or even monocular video) into joint commands for humanoid robots, supporting over 17 robot models and enabling real-time teleoperation and RL training data generation.

libvips is a fast, memory-efficient image processing library with ~300 operations, support for many formats (JPEG, PNG, TIFF, WebP, HEIC, PDF, SVG, etc.), bindings for 11+ languages, and adoption by projects like Mastodon, sharp, and Ruby on Rails. It's LGPL-licensed, installable via package managers or built from source with Meson, and has many optional dependencies for format-specific support.

A macOS command-line tool that lets AI agents draw visible arrows and labels on the screen to guide users to specific UI elements or coordinates.

kimodo.cpp is a C++ GGML implementation of NVIDIA’s Kimodo text‑to‑motion model, offering CPU/Vulkan inference, a C API, and a web demo for generating joint‑level animations from prompts or embeddings.

A command-line tool that generates a standardized, best-practice project template for organizing data science work.

The official command-line interface for Kaggle, allowing users to manage datasets, competitions, models, and notebooks directly from their terminal.

A terminal-based interface for the Kraken exchange that enables manual trading, automated AI-driven trading via MCP, and paper trading with live market data.

A whole-brain leaky integrate-and-fire emulation of the adult fruit fly based on the FlyWire connectome, allowing researchers to simulate and analyze neural spike propagation.

A local-first whiteboard and diagramming application that allows users and AI agents to collaborate on an infinite canvas without requiring an internet connection.

A rule-enforcement tool for AI coding agents that catches and forces repairs for project-specific rule violations that standard linters cannot detect.

A Chinese‑language, hands‑on tutorial that walks users from basic PyTorch concepts to a full GPT‑style implementation, then extends the lessons to modern architectures (GLM‑4, Llama 3, RWKV, MiniCPM). Includes notebooks, training scripts, and exercises, all under Apache 2.0 / CC‑BY‑NC‑SA licensing.

A deployment toolkit for running the GLM-5.3-Flash LLM on NVIDIA DGX Spark clusters, featuring automated orchestration, speculative decoding, and an OpenAI-compatible API.

A modern, asynchronous web scraping and browser automation library for Python that simplifies data extraction, proxy management, and bot-evasion for reliable crawling.

RhinoMCP is an open‑source bridge that lets AI agents control Rhino 3D and Grasshopper via the Model Context Protocol. It exposes high‑level geometry, editing, and Grasshopper‑graph commands as plain‑language tools, enabling AI‑driven design loops.

An agentic AI security tool that identifies exploitable vulnerabilities in source code by simulating attacker paths and using a falsification engine to minimize false positives.

agentUniverse is a multi-agent framework for building intelligent agents powered by large language models. It provides ready-made collaboration patterns (like PEER and DOE) that let agents break down complex tasks, plus easy integration of domain expertise, so developers can quickly create professional-grade AI agents.

An open-source deception framework that deploys realistic, LLM-powered decoys to observe attacker behavior and generate security evidence.

BoxLite is a secure, daemonless compute engine that provides hardware-isolated micro-VMs for AI agents to safely execute code and persist state across tasks.

An on-device AI agent for Android that integrates system-level actions, a local workspace, and an extensible tool ecosystem to execute complex tasks.

ReVa is a Ghidra extension that acts as an MCP server, allowing AI models to perform complex reverse engineering tasks by providing them with a suite of specialized, tool-driven analysis functions.