Top Open Source AI Dev Influencers in May 2025

Get started with building Fullstack Agents using Gemini 2.5 and LangGraph

Collection of extracted System Prompts from popular chatbots like ChatGPT, Claude & Gemini

The simplest, fastest repository for training/finetuning small-sized VLMs.

PandaWiki 是一款 AI 大模型驱动的开源知识库搭建系统，帮助你快速构建智能化的 产品文档、技术文档、FAQ、博客系统，借助大模型的力量为你提供 AI 创作、AI 问答、AI 搜索等能力。

A Model Context Protocol (MCP) Gateway & Registry. Serves as a central management point for tools, resources, and prompts that can be accessed by MCP-compatible LLM applications. Converts REST API endpoints to MCP, composes virtual MCP servers with added security and observability, and converts between protocols (stdio, SSE, Streamable HTTP).

Open-source platform for AI-driven storytelling, worldbuilding, and immersive roleplay

MemoryOS is designed to provide a memory operating system for personalized AI agents.

The world's first Full-Stack Open-Source General AI Agent

AI 洞察日报项目，每日为您精选 AI 领域的最新动态，包括行业新闻、热门开源项目和前沿学术论文以及科技大V推文，并通过 Google Gemini 模型进行智能处理与日报生成，最终自动发布到 GitHub Pages 生成AI日报。

AI-Powered Video Retrieval & Clipping Tool

VLM-3R: Vision-Language Models Augmented with Instruction-Aligned 3D Reconstruction

An example agent demonstrating streaming, tool use, and interactivity from your terminal. This agent builder can help you to build your own agents and tools.

Local LLM Server with GPU and NPU Acceleration

Pixel-Level Reasoning Model trained with RL

Adapting VLMs to Bench2Drive.

CHATS: Combining Human-Aligned Optimization and Test-Time Sampling for Text-to-Image Generation (ICML2025)

AetherLink移动应用是一个基于现代Web技术构建的跨平台AI助手应用。该应用支持与多种AI模型（如OpenAI、Google Gemini、Anthropic Claude、Grok、硅基流动、火山方舟等）的交互，提供流畅的对话体验，并支持Android平台部署。应用采用React、TypeScript和Capacitor框架开发，具有高度可定制的模型配置、多主题聊天管理、AI思考过程可视化、语音合成、语音识别、MCP工具支持、知识库管理等特色功能。

Chat Relay is a system that allows Cline/RooCode to communicate with web-based AI chat interfaces (like Gemini, AI Studio, ChatGPT, and Claude) through an OpenAI-compatible API. This enables using models that may not have public APIs through their web interfaces.

Code for the paper "VTool-R1: VLMs Learn to Think with Images via Reinforcement Learning on Multimodal Tool Use"

Train Large Language Models on MLX.

基于 Google Gemini AI 的全功能图像处理应用

Easiest way to give context to LLMs; Attachments has the ambition to be the general funnel for any files to be transformed into images+text for large language models context by only adding 2 lines to your python code.

Repository for awesome spatial/visual reasoning MLLMs. (focus more on embodied applications)

Github repository for "Bring Reason to Vision: Understanding Perception and Reasoning through Model Merging" (ICML 2025)

Vision Language Models are Biased

An implementation of GRPO for Unsloth's VLMs training

Code2Logic: Game-Code-Driven Data Synthesis for Enhancing VLMs General Reasoning

Take your LLM to the optometrist.

A comprehensive list of awesome research, resources, and tools for leveraging LLMs, VLMs, and VLA models in autonomous driving decision-making and motion planning.

Official Code of IR3D-Bench: Evaluating Vision-Language Model Scene Understanding as Agentic Inverse Rendering

Day-0 support for any Hugging Face model leveraging PyTorch native functionalities while providing performance and memory optimized training and inference recipes.

A platformer game developed in C++ with SDL3, glm, nlohmann-json and Tiled.

A RAG (Retrieval-Augmented Generation) solution Based on Pre-generated QA Pairs. 基于预生成 QA 对的 RAG 知识库解决方案

Llama.cpp runner/swapper and proxy that emulates LMStudio / Ollama backends

cadrille: Multi-modal CAD Reconstruction with Online Reinforcement Learning

Functional tests and benchmarks for verifying Llama model providers.

This is code repo for the paper, Depolying Ten Thousand Robots: Scalable Imitation Learninig for Lifelong Mulit-Agent Path Finding, which won the ICRA 2025 best paper on multi-robot systems and the best student paper.

Awesome Free Platforms for Money Making daily updated by LLM

system prompts of LLMs, AI Tools, AI Products, DeepSeek,ChatGPT, Gemini, Grok, Qwen

Concept for a node-based memory system for LLMs

A Home Assistant custom component that provides an AI-powered agent capable of generating automations based on natural language queries. The agent connects to all entities in your Home Assistant instance and uses OpenAI's or Llama API to translate user requests into valid Home operations including creating automations for you!

A Recipe for Building LLM Reasoners to Solve Complex Instructions

a comprehensive diffusion-based code refinement model

DarkGPT Lite is a specialized CLI tool providing unrestricted conversations with AI for cybersecurity research purposes

A Commitizen plugin that leverages OpenAI's GPT-4o to automatically generate clear, concise, and conventional commit messages based on your staged git changes.

A modern, fast, and beautiful Emacs configuration for developers who want power without the complexity.

VUDA is an autonomous debugging agent that empowers AI models to visually analyze, test, and debug web

Synthetic Dialog Generation and Analysis with LLMs

Everything you need to build, deploy, and collaborate with agents. Ride the llama, avoid the drama.

🧠 Train your own DeepSeek-R1 style reasoning model on Mac! First MLX implementation of GRPO - the breakthrough technique behind R1's o1-matching performance. Build mathematical reasoning AI without expensive RLHF. Apple Silicon optimized. 🚀

Transform any repository into an intelligent code assistant. Upload, analyze, and chat with your codebase using advanced AI with full repository context.

ComfyUI-AniSora is now available in ComfyUI, Index-AniSora is the most powerful open-source animated video generation model. It enables one-click creation of video shots across diverse anime styles including series episodes, Chinese original animations, manga adaptations, VTuber content, anime PVs, mad-style parodies(鬼畜动画), and more!

TIME: A Multi-level Benchmark for Temporal Reasoning of LLMs in Real-World Scenario

ComfyUI custom node powered by LLaVA VLM for advanced image captioning with customizable styles and memory-efficient inference.

A simplified mlx implementation of the original stable diffusion 1.5 algorithm featuring fine-tuned weight loading and lora loading for Text and Image generation

Kuzco is a Swift package for integrating large language models (LLMs) directly into iOS, macOS, and Mac Catalyst apps. Built on `llama.cpp`, it offers customizable prompts, flexible tuning, and async/await-friendly APIs for on-device AI.

在线客服，chatbot，chat，ai，AI Chat，在线客服转人工，大模型聊天，大模型IM客服，可与FreeAICC电话呼叫中心整合。

R1-Zero on any Data

点点现聊，是一款功能丰富、现代化的 Web 聊天应用，它利用 WebRTC 进行用户间的直接媒体通信，支持文本、文件共享、语音消息以及实时音频/视频/屏幕共享通话。通过基于 WebSocket 的信令服务器（由Java Spring Boot实现）完成初始用户发现和连接协商。WebRTC 通讯减少了对中央服务器传输大量媒体数据的依赖（除信令与可选的 TURN 中继服务外），而AI聊天和TTS功能则通过后端代理与外部服务交互。该应用深度集成了具有文本转语音（TTS）功能的主题化AI助手联系人，AI角色拥有动态上下文（如每日随机事件和心情）并支持长对话智能摘要，带来更生动的交互体验。

Lumina is a real-time, multimodal dialogue system that integrates streaming STT, semantic turn detection, memory retrieval, and multi-agent LLM orchestration. Lumina 是一个实时多模态对话系统，集成了流式语音识别、语义轮次检测、记忆检索与多智能体语言模型编排。

[ATC'25] Katz is a high-performance serving system designed specifically for diffusion model workflows with multiple adapters.

Run Llama 3.3, DeepSeek-R1, Phi-4, Gemma 3, Mistral Small 3.1, and other state-of-the-art language models locally with scorching-fast performance. Inferno provides an intuitive CLI and an OpenAI/Ollama-compatible API, putting the inferno of AI innovation directly in your hands.

This repo has specific easy steps for you to be able to train your Llama AI Model offline

Source code for "Towards Efficient Online Tuning of VLM Agents via Counterfactual Soft Reinforcement Learning"

This is a repository for the AgentX project "Improving LLM Agents With Reinforcement Learning on Cryptographic CTF Challenges"

Experiments in giving SDXL a T5 text encoder front end

🔥 The first survey on bridging VLMs and synthetic data, for which I completed the entire process of reading 125 papers and writing the research paper in just 10 days.

Custom ComfyUI nodes used in Maxed Out workflows (SDXL, Flux, etc.)

🤖 A lightweight desktop tool for developers working with AI assistants. 📊 Visualize project structure, 📋 selectively export file paths and code to clipboard, making collaboration with ChatGPT, Claude and other AI assistants more efficient. 🌐 Supports multi-language UI, 🔍 file filtering, and ⚙️ customizable output formats.

🐸 Reddit-Ribbit-Ribbit: A friendly AI Reddit agent that hops into discussions! Powered by Google Gemini 🚀, it crafts engaging, Gen Alpha-style comments with emojis ✨. Features include subreddit monitoring, keyword filtering, image/URL analysis, and a fully customizable "bot brain".

A blog dedicated to helping creators, developers, and curious minds understand the rapidly growing world of Large Language Models (LLMs).

The Aura Emotion AI system has chroma with a local embedding model, memvid qr code mp4 infinite memory, brainwave and neurochemical simulations, sociobiological reasoning, autonomous subsystem processing with a Gemini flash model so the main model is less taxed, is a MCP client with adaptive tool learning and MCP server.

AgentNull is a comprehensive catalog of attack vectors targeting autonomous AI agents, complete with proof-of-concepts for each method. Explore the structured threat information and replicate scenarios using the provided resources. 🐙👨💻

EduNexis is an AI chatbot built specifically for educational purposes. It assists school students—primarily from grades 5 to 10—by answering their academic questions and doubts in a friendly and simple manner. Powered by the LLaMA language model, EduNexis can handle a wide range of topics, subjects, and complexity levels, .

SD/SDXL/Flux Tiny Autoencoder converter for OpenVINO

A simple, intuitive, Roblox admin script.

Convert SD1.5 Textual Inversion to SDXL

This repository offers a well-organized collection of resources focused on reasoning in Large Language Models (LLMs). Explore foundational papers, evaluation benchmarks, and practical tools to enhance your understanding of LLM reasoning. 🐙🌐

An AI development platform that allows users to build custom AI models through an API, supporting chatbots, music generation, and coding tools to streamline creative and technical workflows.

An efficient sampling method for long-CoT LLM with fractured CoT.

This is the Github Repo for the paper: VeriReason: Reinforcement Learning with Testbench Feedback for Reasoning-Enhanced Verilog Generation

https://affective-ai.github.io/

A web-based tool for annotating browser use trajectory for VLMs

AI Assistant that helps you open your Desktop's Apps with voice command

ChatGPT 中文版：国内免费直连指南（支持GPT-4.1、4o画图，无需翻墙）【7月持续更新】

ChatGPT 中文版：国内直连教程（支持GPT-4、4o、o1、o3 和 DeepSeek R1，无需翻墙）【5月持续更新】