Open-source llm-inference software
3358 open-source libraries, tools and applications tagged llm-inference, ranked by how closely llm-inference is their primary purpose, then by adoption and maintenance health. Each entry links to a page with sources, use cases and when to choose or avoid it.
3358 products, primary matches first, then adoption-weighted; health v2 shown.
| Product | Health v2 | Stars | Maturity |
|---|---|---|---|
| Recursive Language Models (RLM) A plug-and-play Python inference library for Recursive Language Models (RLMs), a paradigm where an LLM programmatically examines and decomp… | 76 | 5669 | active |
| YILS-LIN/short-video-factory Short Video Factory is an open-source cross-platform desktop application that uses AI to generate and automatically edit short videos from … | 76 | 5656 | active |
| dotnet/skills A curated collection of agent skills and custom agents from the .NET team, packaged as plugins for AI coding agents like Copilot CLI and Cl… | 63 | 5655 | active |
| nextai-translator/bob-plugin-openai-translator A Bob plugin for macOS that uses OpenAI, Google Gemini, MiniMax, or OpenAI-compatible APIs to translate text, polish wording, and fix gramm… | 93 | 5651 | active |
| zarazhangrui/codebase-to-course A Claude Code skill that converts any codebase into a self-contained, interactive single-page HTML course for non-technical 'vibe coders'. … | 45 | 5642 | active |
| ConnectAI-E/feishu-openai A self-hosted Go application that integrates OpenAI models (GPT-4, GPT-4V, DALL·E-3, Whisper) into Feishu/Lark as a chatbot. It supports vo… | 32 | 5633 | active |
| HisMax/RedInk RedInk is a self-hostable web application that generates complete Xiaohongshu (Little Red Book) image-and-text posts from a single topic se… | 54 | 5632 | active |
| lm-sys/RouteLLM RouteLLM is a Python framework for serving and evaluating LLM routers that route queries between strong and weak models to cut costs. It pr… | 25 | 5626 | active |
| maziyarpanahi/openmed OpenMed is a local-first healthcare AI SDK for clinical named-entity recognition and HIPAA PII/PHI de-identification that runs entirely on-… | 84 | 5623 | active |
| sohzm/cheating-daddy Cheating Daddy is a free, open-source Electron desktop app that acts as a real-time AI assistant during video calls, interviews, and meetin… | 72 | 5616 | active |
| workweave/router Workweave Router is a model router for agentic AI systems that routes each prompt to the most cost-effective model capable of handling it, … | 66 | 5596 | active |
| MiniGPT-4 Official code for MiniGPT-4 and MiniGPT-v2, vision-language models that align a frozen visual encoder with a frozen LLM (Vicuna) via a sing… | 30 | 25605 | maintenance |
| karpathy/llm-council A local web app that sends a user's question to multiple LLMs via OpenRouter, has them anonymously review and rank each other's answers, an… | 37 | 25563 | maintenance |
| ngxson/smolvlm-realtime-webcam A browser-based demo that streams webcam frames to a llama.cpp server running SmolVLM 500M for real-time object detection and scene descrip… | 26 | 5577 | active |
| ATH-MaaS/ComfyUI-Copilot ComfyUI-Copilot is an AI-powered custom node for ComfyUI that acts as an intelligent assistant for building, debugging, and optimizing imag… | 61 | 5539 | active |
| father-bot/chatgpt_telegram_bot A self-hostable Telegram bot that brings ChatGPT and Claude models into Telegram using your own OpenAI/Anthropic/OpenRouter API keys. It su… | 73 | 5538 | active |
| ysharma3501/LuxTTS LuxTTS is a lightweight zipvoice-based text-to-speech model for high-quality zero-shot voice cloning, generating clear 48kHz speech at up t… | 51 | 5433 | active |
| docker/genai-stack A Docker Compose-based starter stack for building GenAI applications, combining LangChain, Ollama for local LLM inference, and Neo4j as a d… | 70 | 5416 | active |
| EverMind-AI/Raven Raven is an open-source, self-improving agent harness built on EverOS that combines terminal-first execution, local tracing, long-term memo… | 81 | 5409 | active |
| TaskingAI/TaskingAI TaskingAI is an open-source Backend-as-a-Service (BaaS) platform for developing and deploying LLM-based AI agents. It unifies access to hun… | 17 | 5406 | active |
| VibeSDK An open-source SDK for building your own vibe-coding platform where users describe full-stack apps and an AI coding agent plans, edits, dep… | 71 | 5401 | active |
| lingodotdev/lingo.dev Lingo.dev is an open-source localization engineering toolkit that turns LLMs into stateful translation APIs, persisting glossaries, brand v… | 88 | 5398 | active |
| NVIDIA/Model-Optimizer NVIDIA Model Optimizer (ModelOpt) is a Python library of state-of-the-art model optimization techniques including quantization, pruning, di… | 92 | 5386 | active |
| deepseek-ai/DeepSeek-VL2 DeepSeek-VL2 is a series of Mixture-of-Experts vision-language models (Tiny, Small, and 4.5B activated parameters) with inference code and … | 22 | 5376 | active |
| looplj/axonhub AxonHub is an open-source AI gateway written in Go that lets any SDK call 100+ LLMs through a unified API. It provides built-in failover, l… | 81 | 5375 | active |
| lmstudio-ai/lms lms is the official command-line tool for LM Studio, built on the lmstudio.js SDK. It lets users manage local LLMs (list, download, load, u… | 70 | 5368 | active |
| winfunc/deepreasoning DeepReasoning is a high-performance Rust-based LLM inference API and chat UI that combines DeepSeek R1's chain-of-thought reasoning traces … | 28 | 5361 | active |
| Roo Code Roo Code is a VS Code extension that brings a team of AI agents into the editor to assist with coding, design, debugging, and documentation… | 10 | 24296 | maintenance |
| aipoch/open-science Open Science is an open-source, local-first, model-agnostic AI research workbench for reproducible scientific research. It combines a coord… | 77 | 5303 | active |
| BuilderIO/ai-shell AI Shell is an open-source CLI tool that converts natural language prompts into shell commands using OpenAI, with options to run, revise, o… | 54 | 5301 | active |
| neo4j-labs/llm-graph-builder A web application that transforms unstructured data (PDFs, DOCs, TXTs, YouTube videos, web pages) into a knowledge graph stored in Neo4j us… | 88 | 5264 | active |
| Anil-matcha/AI-Youtube-Shorts-Generator An open-source Python tool that turns long-form YouTube videos into vertical 9:16 short clips using LLM-based highlight detection, Whisper … | 68 | 5262 | active |
| camel-ai/oasis OASIS is an open-source social media simulator that uses large language model and rule-based agents to mimic the behavior of up to one mill… | 72 | 5245 | active |
| su-kaka/gcli2api A self-hosted Python service that converts GeminiCLI and Antigravity into OpenAI-, Gemini-, and Claude-compatible API endpoints. It can be … | 63 | 5243 | active |
| DeepResearch A TypeScript agent that iteratively searches the web, reads webpages, and reasons with an LLM (Gemini, OpenAI, or local) until it finds a c… | 55 | 5237 | active |
| pollinations/pollinations Pollinations is an open-source generative AI platform offering free APIs for text, image, and vision model inference without requiring API … | 77 | 5226 | active |
| Kode AI Agent Platform Learn Claude Code is a Python library that provides a minimal harness for building Claude-like AI agents. It includes CLI tools and tutoria… | 60 | 5225 | active |
| Adam-CAD/CADAM CADAM is an open-source web application that converts natural language descriptions and images into 3D CAD models using LLM-powered agents … | 73 | 5210 | active |
| linuxhsj/openclaw-zero-token OpenClaw Zero Token is a fork of OpenClaw that provides a unified gateway to major LLM providers (ChatGPT, Claude, Gemini, DeepSeek, Qwen, … | 71 | 5206 | active |
| kodu-ai/claude-coder Kodu (Claude Coder) is an autonomous AI coding agent distributed as a VS Code extension. It takes natural-language task descriptions and bu… | 28 | 5205 | active |
| Transformer Lab Transformer Lab is an open-source desktop application (built with Electron and Python) that provides a unified GUI for training, fine-tunin… | 82 | 5195 | active |
| AgnesAI-Labs/AgnesAI-Models Official gateway and model catalog for Agnes AI, providing OpenAI-compatible API access to multimodal foundation models covering text, imag… | 55 | 5192 | active |
| vllm-project/aibrix AIBrix is an open-source, cloud-native framework from the vLLM project providing essential building blocks for scalable, cost-efficient Gen… | 88 | 5149 | active |
| Codeium/Windsurf windsurf.vim is a Vim and Neovim plugin providing free, AI-powered code autocomplete as a Copilot alternative. It integrates the Windsurf (… | 62 | 5142 | active |
| OpenBMB/AgentVerse AgentVerse is a Python framework for deploying multiple LLM-based agents, offering two frameworks: task-solving (multi-agent collaboration … | 21 | 5141 | active |
| icip-cas/PPTAgent PPTAgent is an agentic Python framework that automatically generates PowerPoint presentations from documents using a two-stage, edit-based … | 79 | 5122 | active |
| tencentmusic/supersonic SuperSonic is an AI+BI platform that combines Chat BI (natural language querying powered by LLMs) with Headless BI (a governed semantic lay… | 62 | 5112 | active |
| ztxz16/fastllm fastllm is a high-performance C++ LLM inference library with its own custom operators, requiring no PyTorch dependency. It supports tensor-… | 75 | 5110 | active |
| pytorch/executorch ExecuTorch is PyTorch's framework for exporting and running AI models on-device across mobile, embedded, and edge hardware, with a tiny (~5… | 94 | 5101 | active |
| manycore-research/SpatialLM SpatialLM is a 3D large language model that processes point cloud data (from monocular video, RGBD images, or LiDAR) and generates structur… | 59 | 5094 | active |
| huggingface/text-embeddings-inference Text Embeddings Inference (TEI) is a Rust-based toolkit from Hugging Face for deploying and serving open-source text embedding, re-ranker, … | 93 | 5079 | active |
| deepseek-ai/DeepSeek-V2 DeepSeek-V2 is a strong, economical Mixture-of-Experts language model released with open weights, inference code, and evaluation tooling. T… | 25 | 5042 | active |
| unicity-aos/capsule-react A Rust ReAct loop coordinator that implements a stateless state machine for the reasoning-and-action cycle in agent systems. It is part of … | 70 | 5041 | active |
| developersdigest/llm-answer-engine A Perplexity-inspired AI answer engine built with Next.js, the Vercel AI SDK, LangChain.JS, Groq/Mixtral, Brave Search, Serper, and OpenAI.… | 58 | 5035 | active |
| Gaia node GaiaNet node is a shell-based installer and runtime for running your own decentralized AI agent service locally. It deploys an LLM inferenc… | 45 | 5015 | active |
| AG2 AG2 (formerly AutoGen) is an open-source Python framework for building AI agents and orchestrating multi-agent systems, supporting LLM inte… | 85 | 5003 | active |
| pkuliyi2015/multidiffusion-upscaler-for-automatic1111 An Automatic1111 Stable Diffusion WebUI extension that enables generating and upscaling ultra-large images (2K+) on GPUs with limited VRAM … | 31 | 4995 | active |
| gd3kr/BlenderGPT A Blender add-on that integrates OpenAI's GPT-4/GPT-3.5 into Blender's UI, letting users control the 3D software with natural language comm… | 31 | 4987 | active |
| mistralai/mistral-vibe Mistral Vibe is an open-source command-line coding agent powered by Mistral's models, providing a conversational interface to explore and m… | 77 | 4986 | active |
| going-doer/Paper2Code PaperCoder is a multi-agent LLM system that automatically generates a code repository from a machine learning research paper via a planning… | 49 | 4982 | active |
| EvoScientist EvoScientist is a Python-based multi-agent framework for building self-evolving AI scientists that plan, research, code, analyze data, and … | 83 | 4980 | active |
| joonspk-research/generative_agents The official code release for the Stanford 'Generative Agents' research paper, implementing computational agents powered by LLMs that simul… | 29 | 22237 | maintenance |
| elie222/rakazo Rakazo is an open-source, self-hosted platform for running persistent AI teammates (bots) that can use a browser, terminal, and desktop, re… | 81 | 4947 | active |
| google-deepmind/alphageometry Google DeepMind's implementation of AlphaGeometry and DDAR, AI systems that solve Olympiad-level geometry theorem proving problems. It comb… | 52 | 4899 | active |
| Martian-Engineering/lossless-claw A Lossless Context Management (LCM) plugin for OpenClaw that replaces sliding-window context truncation with a DAG-based summarization syst… | 79 | 4893 | active |
| llm-d llm-d is a Kubernetes-native distributed LLM inference serving stack that orchestrates model servers like vLLM and SGLang across clusters. … | 83 | 4891 | active |
| Deep Java Library (DJL) Deep Java Library (DJL) is an engine-agnostic, high-level deep learning framework for Java that supports backends like PyTorch, TensorFlow,… | 96 | 4857 | active |
| chaitin/MonkeyCode MonkeyCode is an open-source, enterprise-grade AI coding platform that combines managed development environments, AI model management, and … | 78 | 4811 | active |
| SakuraLLM/SakuraLLM SakuraLLM is a series of fine-tuned open large language models for Japanese-to-Chinese translation specialized in light novel and Galgame (… | 58 | 4785 | active |
| Integuru-AI/Integuru Integuru is an AI agent that reverse-engineers platforms' internal APIs by analyzing browser network requests (HAR files) and building depe… | 59 | 4781 | active |
| facebookresearch/lingua Meta Lingua is a minimal, fast LLM training and inference library built on easy-to-modify PyTorch components for research purposes. It supp… | 33 | 4771 | active |
| deepseek-ai/Engram Official implementation of Engram, a conditional memory module from DeepSeek that modernizes N-gram embeddings for O(1) lookup as a new spa… | 41 | 4735 | active |
| GeekAI GeekAI is an open-source, self-hosted AI content creation platform integrating chat, image generation (MidJourney, DALL-E, Stable Diffusion… | 92 | 4709 | active |
| Rivet Rivet is an open-source visual AI programming environment consisting of a desktop IDE for building AI agents and prompt chains via a node-b… | 73 | 4708 | active |
| u14app/deep-research Deep Research is a web application that uses any large language model (Gemini, OpenAI, Anthropic, DeepSeek, Grok, Ollama, etc.) combined wi… | 66 | 4691 | active |
| OpenNMT/CTranslate2 CTranslate2 is a C++ and Python library for fast, memory-efficient inference of Transformer models on CPU and GPU. It uses quantization, la… | 94 | 4684 | stable |
| OpenMinis OpenMinis is a free, open-source mobile AI agent app for iOS and Android that connects leading LLM providers (Claude, GPT, Gemini, etc.) vi… | 76 | 4673 | active |
| Tencent/TNN TNN is a high-performance, lightweight deep learning inference framework developed by Tencent Youtu Lab, supporting mobile, desktop, and se… | 28 | 4656 | active |
| ruc-datalab/DeepAnalyze DeepAnalyze is the first agentic LLM (DeepAnalyze-8B, built on Qwen) for autonomous data science, capable of running the full pipeline from… | 61 | 4655 | active |
| promptslab/Promptify Promptify is a Python library that turns LLMs into task-based NLP engines, providing high-level APIs for NER, classification, QA, and custo… | 62 | 4636 | active |
| TencentCloudADP/youtu-agent Youtu-Agent is a Python framework for building, running, and evaluating autonomous LLM agents, supporting ReAct-style single agents and pla… | 52 | 4622 | active |
| JetBrains/koog Koog is a JetBrains-built Kotlin/Java framework for creating fault-tolerant, enterprise-ready AI agents on the JVM, with multiplatform depl… | 86 | 4617 | active |
| hao-ai-lab/FastVideo FastVideo is a unified Python framework for post-training and real-time inference of video diffusion models, covering data preprocessing, f… | 79 | 4616 | active |
| mini-swe-agent SWE-agent is a Python framework that lets a language model autonomously use tools to fix GitHub issues, find cybersecurity vulnerabilities,… | 67 | 20450 | maintenance |
| VRSEN/agency-swarm Agency Swarm is a Python framework for building multi-agent AI applications on top of the OpenAI Agents SDK. It lets developers define spec… | 91 | 4601 | active |
| FareedKhan-dev/all-agentic-architectures A Python library packaging 35 production-grade agentic AI architectures (Reflexion, LATS, GraphRAG, MemGPT, Voyager, etc.) as runnable Arch… | 57 | 4600 | active |
| get-convex/chef Chef is an AI-powered full-stack web app builder by the Convex team, forked from bolt.diy, that generates apps with a built-in reactive dat… | 55 | 4599 | active |
| kyegomez/tree-of-thoughts A Python library implementing the Tree of Thoughts algorithm for deliberate, multi-step problem solving with large language models. It prov… | 64 | 4595 | active |
| MemeCalculate/moyin-creator Moyin Creator is a production-grade AI film creation desktop application that covers the full pipeline from script to finished video, with … | 73 | 4580 | active |
| fixie-ai/ultravox Ultravox is a fast multimodal LLM that understands human speech directly without a separate ASR stage, by projecting audio into the LLM's e… | 40 | 4577 | active |
| phodal/auto-dev AutoDev is an AI-native multi-agent coding development platform built on Kotlin Multiplatform, distributed as an IntelliJ IDEA plugin, VS C… | 80 | 4548 | active |
| 0xNyk/council-of-high-intelligence A shell-based skill/plugin that runs structured multi-persona deliberation for hard decisions across AI coding agents like Claude Code, Cod… | 72 | 4539 | active |
| OAID/Tengine Tengine is a lightweight, high-performance, modular deep learning inference engine developed by OPEN AI LAB for embedded and edge devices. … | 24 | 4531 | active |
| kaixindelele/ChatPaper ChatPaper is a Python tool that uses ChatGPT to summarize arXiv papers, translate full PDFs, polish academic writing, and simulate peer rev… | 59 | 19943 | maintenance |
| Embabel Agent Framework Embabel Agent is an agentic AI framework for the JVM, written in Kotlin with first-class Java and Spring Boot support. It provides typed, g… | 87 | 4513 | active |
| openai/harmony OpenAI's Rust library (with first-class Python bindings) for rendering and parsing the harmony response format used by its open-weight gpt-… | 49 | 4506 | active |
| microsoft/LMOps A Microsoft research initiative collecting code and papers on fundamental techniques for building AI products with large language models, c… | 75 | 4475 | active |
| crmne/ruby_llm RubyLLM is a Ruby framework providing a unified, expressive interface to all major AI providers (OpenAI, Anthropic, Google, Ollama, and any… | 88 | 4472 | active |
| anc95/ChatGPT-CodeReview A GitHub App and GitHub Action that automatically reviews pull requests using ChatGPT/OpenAI models, posting AI-generated review comments o… | 89 | 4466 | active |
| EvolvingLMMs-Lab/lmms-eval lmms-eval is a unified Python framework for evaluating large multimodal models across text, image, video, and audio tasks, with 100+ benchm… | 87 | 4453 | active |