Open-Source AI · Tools

Open-Source AI Tools

A curated directory of 232 of the best open-source AI tools in 2026 — each with a deep-dive, alternatives and head-to-head comparisons. Search or filter by category.

Curated by OpenSourceAI.tech

Run LLMs locally

OllamaRun open LLMs locally from one commandLM StudioBrowse, download and chat with local modelsJanOpen-source, offline ChatGPT-style desktop appGPT4AllPrivate local AI that runs on CPUllama.cppThe C/C++ engine powering local inferenceLocalAIA drop-in OpenAI API you self-hostText Generation WebUIFeature-rich web UI for local modelsKoboldCppSingle-file local model runnerMLC LLMRun LLMs on any device, even phonesllamafileOne executable file = model + runtimeexoRun big models across your everyday devicesCortexOllama-style runtime from the Jan teamNexa SDKRun any model on any device — CPU, GPU, NPURamaLamaRun models as OCI containersGPUStackManage GPU clusters for running models

Self-hosted chat UI

Open WebUIChatGPT-style interface over any modelLibreChatMulti-provider open-source ChatGPT cloneLobe ChatModern, design-forward chat frameworkAnythingLLMAll-in-one app to chat with your documentsbig-AGIPro-grade chat for many modelsChatboxDesktop client for many AI modelsNextChatLightweight ChatGPT-style UI, deploys in one clickChatbot UIClean open ChatGPT clone with files and profilesHuggingChat UIThe interface behind HuggingChat, self-hostableKhojYour AI second brain over your notesKotaemonClean open UI to chat with documentsOpen WebUI PipelinesAdd custom logic to any OpenAI-compatible UIMstyOffline chat with split-screen model comparison

AI image generation

ComfyUINode-graph control over image pipelinesAUTOMATIC1111The classic Stable Diffusion web UIFooocusMidjourney-like simplicity, zero tweakingInvokeAIPolished studio with a unified canvasSD.NextPower-user fork with broad model supportSD WebUI ForgeA faster Stable Diffusion WebUISwarmUIFriendly front-end with ComfyUI power insideKrita AI DiffusionGenerative AI inside a real painting appStability MatrixOne app to install and manage every SD interfaceFluxGymTrain image LoRAs with a simple web UIOneTrainerOne tool to fine-tune any diffusion modelIP-AdapterUse an image as a prompt

Inference server

vLLMHigh-throughput serving for productionTGIHugging Face's production text serverSGLangFast serving with structured outputsLMDeployToolkit for compressing and serving LLMsAphrodite EngineHigh-throughput LLM servingTensorRT-LLMPeak throughput on NVIDIA GPUsOpenLLMServe any open model as an OpenAI API in one commandKTransformersRun huge MoE models on one consumer GPURay ServeScale model serving across a clusterBentoMLPackage any model into a production API

Coding assistant

ContinueOpen-source autocomplete and chat in your IDEAiderAI pair programming in your terminalClineAutonomous coding agent in VS CodeRoo CodeCustomizable agentic coder (Cline fork)TabbySelf-hosted Copilot alternativeGooseOn-machine AI agent for engineering tasksOpenCodeOpen Claude-Code-style agent for your terminalVoidOpen-source Cursor alternative on VS CodeZedBlazing-fast editor with native AI and collabPlandexPlan-first agent for large codebasesSourcegraph CodyCodebase-aware AI assistantGPT EngineerGenerate a whole codebase from a promptSweepTurn issues into pull requests

LLM / RAG framework

LangChainCompose chains, tools and agentsLlamaIndexThe data framework for RAGHaystackProduction pipelines for search and RAGDSPyProgram — not prompt — language modelstxtaiAll-in-one embeddings databaseRAGFlowDeep-document-understanding RAGSemantic KernelMicrosoft's enterprise agent frameworkGraphRAGRAG that builds a knowledge graph firstLiteLLMOne API for 100+ LLM providersInstructorReliable structured outputs from LLMsOutlinesGuarantee valid JSON and grammarsGuidanceInterleave control and generationLLMWareEnterprise RAG with small specialised modelsFlashRankUltra-light reranking for better RAGSentence TransformersThe standard way to make embeddingsRagasMeasure whether your RAG is any goodPromptfooUnit tests for your promptsLangfuseSee what your LLM app actually didPhoenixTrace, evaluate and debug LLM apps

AI agent framework

CrewAIOrchestrate role-playing agent crewsPaperclipRun a company of AI agents from one dashboardAutoGenMicrosoft's conversational agent frameworkLangGraphStateful, controllable agent graphsMetaGPTA software company of LLM agentsAutoGPTThe original autonomous GPT agentOpenHandsAI developer that writes and runs codeSuperAGIDev-first autonomous agent platformBabyAGIMinimal task-loop autonomous agentAgnoFast, lightweight multi-modal agentsLettaStateful agents with long-term memorysmolagentsAgents in a few lines of codePydantic AIType-safe agents for productionSWE-agentAgent that fixes GitHub issuesAgent ZeroA general agent that grows with your promptsCAMELThe research framework for agent societiesElizaAgents with personality for Discord, X and moreGoogle ADKGoogle's official Agent Development KitMem0Long-term memory for AI agentsComposio250+ ready-made tools for your agentsAG2The community continuation of AutoGenOpenAI SwarmMinimal multi-agent handoffsPhidataAgents with memory, knowledge and toolsJulepStateful agents as a service

Browser & computer-use agents

Browser UseLet AI agents control your browserSkyvernAutomate browser workflows with LLMs + visionStagehandAI browser automation on top of PlaywrightOpen InterpreterChatGPT's Code Interpreter on your machineNanobrowserMulti-agent web automation in your own browserSteelHeadless browser infrastructure for AI agentsPlaywright MCPGive any AI a real browser via MCPLaVagueTurn objectives into browser automation

Low-code AI builder

DifyBuild and operate AI apps visuallyFlowiseDrag-and-drop LLM flowsLangflowVisual builder for agents and RAGn8nAutomate workflows with AI nodesActivepiecesOpen-source Zapier with AI piecesWindmillTurn scripts into workflows and UIsKestraDeclarative orchestration with AI steps

Vector database

QdrantFast Rust-based vector searchWeaviateVector DB with built-in modulesMilvusBillion-scale vector databaseChromaThe simplest embedding DB for prototypingpgvectorVector search inside PostgreSQLLanceDBServerless vector DB for AIFAISSThe reference library for similarity searchVespaBig-scale hybrid search and ranking platformpgvectorscaleMake pgvector fast at scaleMarqoVector search with embedding built inUSearchTiny, extremely fast vector index

Speech (STT / TTS)

WhisperOpenAI's open speech-to-text baselinefaster-whisperWhisper, much faster and lighterWhisperXWhisper plus word timestamps and diarizationPiperFast, local neural text-to-speechCoqui TTSDeep-learning TTS with voice cloningBarkText-to-audio with voices and effectsF5-TTSZero-shot voice cloning that actually convincesKokoroTiny 82M TTS with astonishing qualitywhisper.cppWhisper in pure C/C++, runs anywhereOpenVoiceClone a voice and control its emotionStyleTTS 2Human-level speech synthesispyannote.audioKnow who spoke whenSilero VADDetect speech in audio, instantly

Fine-tuning

UnslothFine-tune LLMs 2x faster on one GPUAxolotlConfig-driven fine-tuning for many modelsLLaMA-FactoryFine-tune 100+ models with a UIPEFTLoRA and friends from Hugging FaceTRLAlign LLMs (SFT, DPO, PPO)TorchtunePyTorch-native post-training, hackable recipesLlama CookbookOfficial recipes to fine-tune Llamams-swiftFine-tune 500+ LLMs and VLMs

Document AI & OCR

MarkerConvert PDFs to clean Markdown fastDoclingIBM-grade document understandingSuryaModern OCR with layout detectionPaddleOCRIndustrial OCR with tiny modelsUnstructuredTurn any file into LLM-ready dataFirecrawlTurn websites into LLM-ready MarkdownCrawl4AIOpen-source LLM-friendly crawlerolmOCRTurn PDFs into clean training-grade textMinerUPDF to Markdown with formulas and tablesTesseractThe classic OCR engine

AI video generation

Open-SoraOpen video generation you can runWan2.1State-of-the-art open video modelHunyuanVideoTencent's large open video modelLTX-VideoReal-time-class video generationFramePackLong video on modest VRAMMochi 1High-fidelity open video generationCogVideoText and image to videoAnimateDiffAnimate any Stable Diffusion model

AI search engines

PerplexicaOpen-source Perplexity you self-hostMorphicGenerative UI answer engineSearXNGThe private search layer behind AI searchFarfalleSelf-hosted answer engine with local modelsLepton SearchA conversational search engine in 500 linesSurfSenseSearch your own knowledge and the web

Learn AI & machine learning

VirgilioA structured mentor for data science and MLML for BeginnersMicrosoft's classic machine learning courseAI for BeginnersFrom symbolic AI to neural networksGenerative AI for BeginnersBuild generative AI apps, lesson by lessonData Science for BeginnersThe data foundations before any MLLLM CourseThe reference roadmap for learning LLMsLLMs from ScratchBuild a GPT from nothing, line by lineNeural Networks: Zero to HeroKarpathy builds backprop, then GPT, from scratchAnnotated Paper Implementations60+ papers implemented and explained side by sideDive into Deep LearningThe textbook where every equation is runnablefastai Book (fastbook)Deep learning for coders, top-downHands-On Machine LearningThe notebooks of the best-selling ML bookHugging Face CourseMaster transformers with the actual libraryHugging Face Agents CourseBuild AI agents, properlyMade With MLFrom notebook to production systemPrompt Engineering GuideThe reference on prompting, backed by papersApplied MLHow real companies actually ship MLML YouTube CoursesThe best free ML courses on YouTube, curatedOpenAI CookbookPractical recipes that work with any OpenAI-compatible APIAwesome Machine LearningThe reference index of ML libraries, by languageAwesome LLMPapers, models and tools of the LLM eraDeep Learning DrizzleUniversity lectures, from the sourceML Interviews BookWhat ML interviews actually ask

Robotics & embodied AI

GenesisGenerate robotic worlds from a text promptLeRobotTrain real robots with the Hugging Face stackArduPilotAutopilot for drones, rovers and boatsMuJoCoThe physics engine most robotics research runs onStable-Baselines3Reliable RL algorithms you can actually trustopenpi (π0)Open weights for robot foundation modelsGymnasiumThe standard interface for reinforcement learningPX4The autopilot behind most commercial dronesAutowareA complete open-source self-driving stackIsaac LabMassively parallel robot training on NVIDIA GPUsROS 2The operating system layer of modern roboticsNav2Make a mobile robot navigate on its ownDiffusion PolicyTeach a robot by showing it, using diffusionHabitatTrain agents to act in photorealistic homesMoveIt 2Motion planning and manipulation for robot armsGazeboSimulate a whole robot, sensors included

ML frameworks & MLOps

DagsterOrchestration that thinks in data assets, not tasksTensorFlowGoogle's deep-learning framework, built for productionPyTorchThe framework nearly every modern AI model is written inOpenCVThe computer vision library everything else builds onscikit-learnClassical machine learning, done properlyApache AirflowSchedule and monitor data pipelinesRayScale Python from a laptop to a clusterJAXNumPy with autodiff, JIT and TPUsXGBoostStill the one to beat on tabular dataLabel StudioLabel anything — text, images, audio, videoMLflowTrack experiments and ship models without the spreadsheetONNXMove a model between frameworks and runtimesLightGBMGradient boosting that trains fast on big tablesCVATSerious annotation for computer visionDVCGit for datasets and modelsOptunaFind the right hyperparameters without guessing

Compare them head-to-head

See detailed side-by-side comparisons of these tools, with feature matrices and scores.

Open the comparisons →