Open-source tools
Open-source and local-first AI tools.
This page separates open-source, open-core, and source-available tools from commercial workflow listings so builders can focus on local AI, RAG, inference, coding agents, evaluation, and infrastructure.
Reviewed June 2026
How to use this page
Treat this as the open-source lane of the tools directory. Review each tool profile for licensing, deployment notes, maintenance requirements, alternatives, and links to official documentation.
60 open-source and open-core tools
Coding agent
Cline
Open VS Code coding agent for reviewable file edits, configurable model providers, and agentic coding workflows.
Best for: Developers who want a more controllable AI coding agent workflow.
Tool details →Local runner
Ollama
Run open models locally with a simple CLI, model library, desktop app, and local API.
Best for: Developers who want a single command to download, run, and serve any GGUF or safetensors model on local hardware — with automatic GPU layer allocation across CUDA, Metal, and ROCm backends — as an OpenAI-compatible API for any downstream tool.
Tool details →Local runner
Jan
Open-source desktop app for running local AI models with a friendly ChatGPT-like interface.
Best for: Privacy-conscious developers and individuals who want a fully open-source desktop AI workspace — local model download, chat, and API server — as an auditable alternative to proprietary tools like LM Studio.
Tool details →Desktop client
Cherry Studio
Open-source desktop AI client that puts local runtimes, remote provider APIs, MCP servers, and document knowledge bases behind one chat interface.
Best for: People who want one desktop chat client for both local and hosted models, and who want MCP tool access without wiring up a separate agent framework.
Tool details →Local runner
GPT4All
Local AI desktop and SDK project for running open models on consumer machines.
Best for: Non-technical users and teams evaluating private local LLM chat who want the simplest possible setup — double-click install, model download, and immediate chat — without any terminal, Docker, or API configuration.
Tool details →Local runner
llamafile
Mozilla-backed project for packaging LLMs into portable executable files.
Best for: Developers who need to distribute a model plus runtime as a single executable artifact — for demos, air-gapped deployments, conference presentations, or testing environments where zero-installation portability matters more than multi-model management.
Tool details →Local runner
llama.cpp
Core C/C++ inference project behind many local GGUF model workflows.
Best for: Engineers running quantized models on CPU-only or mixed CPU/GPU hardware who need direct control over inference parameters, memory layout, and runtime flags.
Tool details →Chat workspace
Open WebUI
Self-hosted AI platform and chat interface for Ollama and other model backends.
Best for: Teams and individuals who want a full-featured self-hosted chat interface connected to local Ollama models or OpenAI-compatible endpoints, without sending conversation data to external services.
Tool details →Chat workspace
AnythingLLM
Open-source AI workspace for document chat, agents, and local or hosted models across desktop, self-hosted, and cloud setups.
Best for: Teams and individuals who want private document chat and multi-user AI workspaces without locking into one model provider.
Tool details →Chat workspace
LibreChat
Open-source multi-provider chat platform with a familiar assistant interface.
Best for: Teams that want a feature-rich, self-hosted chat interface supporting simultaneous access to multiple LLM providers — OpenAI, Anthropic, Google, Azure, and local Ollama endpoints — in a ChatGPT-familiar UI without a monthly subscription.
Tool details →Chat workspace
LobeChat
Open-source modern chat UI for multiple model providers and assistant workflows.
Best for: Builders and internal tools teams who want a polished, consumer-grade chat UI with multi-provider support and plugin extensibility that deploys to Vercel or Docker in minutes without building UI from scratch.
Tool details →Coding assistant
Continue
Open-source AI coding assistant for VS Code and JetBrains with configurable models and context.
Best for: Development teams who want a model-agnostic coding assistant in their existing IDE with source-controlled configuration — connecting local Ollama models, self-hosted endpoints, or cloud APIs through a single versioned config file.
Tool details →Coding agent
Aider
Terminal-based AI pair programmer that edits files in your local git repo.
Best for: Developers who want AI-driven file edits automatically tracked as reviewable git diffs — using their preferred model via API or local Ollama endpoint — without leaving the terminal.
Tool details →Coding agent
Cline
Open-source VS Code coding agent for planning, editing files, and using tools.
Best for: Developers who want an autonomous coding agent inside VS Code that can read files, execute shell commands, search the web, and call MCP servers — with explicit human approval checkpoints before any filesystem or process action.
Tool details →Coding agent
Roo Code
VS Code AI coding agent forked from the Cline ecosystem with multi-mode workflows.
Best for: Developers who want configurable agent roles — Code, Architect, Ask, Debug — each with independent model assignments and tool permissions, enabling cost optimization across different task complexity tiers.
Tool details →Coding agent
Kilo Code
Open-source VS Code coding agent focused on agentic development workflows.
Best for: VS Code developers evaluating actively maintained open-source coding agents in the Cline ecosystem who need agentic file editing with MCP tool integration and regular upstream maintenance commits.
Tool details →Coding agent
OpenCode
Terminal-based coding agent for working with models and repositories from the command line.
Best for: Developers who prefer terminal-native coding agent workflows and want a TUI-driven alternative to Aider with a different approach to session management and model interaction.
Tool details →Coding assistant
Tabby
Self-hosted AI coding assistant for code completion and team-controlled coding workflows.
Best for: Engineering teams that need self-hosted code completion infrastructure with organization-level control over model choice, data residency, and usage analytics — without sending proprietary code to external cloud services.
Tool details →RAG and workflow
Flowise
Open-source visual builder for AI agents, chat assistants, RAG flows, and multi-agent systems.
Best for: Builders who want to prototype and deploy LLM chatbots, RAG flows, and agent workflows visually before committing to custom code — particularly teams evaluating architecture feasibility before engineering investment.
Tool details →RAG and workflow
Langflow
Open-source visual Python platform for building AI applications, agents, RAG flows, APIs, and MCP tools.
Best for: Python-oriented teams building visual RAG, agent, data, API, or MCP workflows that still need custom-code extensibility.
Tool details →RAG and workflow
LangChain
Framework ecosystem for LLM apps, agents, tools, retrieval, and observability workflows.
Best for: Python developers building custom LLM applications that need broad integration coverage — connecting document loaders, vector stores, external APIs, and model providers without writing every connector from scratch.
Tool details →RAG and workflow
LlamaIndex
Data framework for connecting LLMs to documents, databases, retrieval, and agents.
Best for: Developers building data-intensive RAG systems where document ingestion, chunking strategy, retrieval quality, and pipeline evaluation are central engineering concerns rather than afterthoughts.
Tool details →RAG and workflow
Haystack
Open-source framework for production-style search, RAG, and NLP pipelines.
Best for: Engineering teams building production-grade retrieval and NLP pipelines where composable pipeline architecture, component-level testing, and explicit data flow contracts are as important as raw functionality.
Tool details →Agent framework
CrewAI
Framework for orchestrating role-based AI agents and multi-agent workflows.
Best for: Developers building multi-step AI workflows where distinct agent roles — researcher, writer, reviewer, executor — collaborate around a shared task with defined handoff points and expected output schemas.
Tool details →Agent framework
AutoGen
Open-source programming framework from Microsoft for building agentic and multi-agent AI systems.
Best for: Developers and researchers building multi-agent systems where specialized AI agents need to collaborate, debate, and hand off tasks through structured conversational state machines rather than simple sequential prompt chains.
Tool details →Automation
n8n
Workflow automation platform for connecting AI tools, APIs, databases, and business systems.
Best for: Builders who want to connect AI model calls, webhooks, business APIs, and data operations into automated workflows without writing custom integration code — particularly where AI is one node in a larger business process, not the entire stack.
Tool details →Vector database
Qdrant
Open-source vector database with strong filtering and production-oriented retrieval features.
Best for: Teams building production RAG systems that need precise metadata filtering alongside vector similarity search — filtering on document source, date range, user permissions, or content type while maintaining sub-100ms retrieval latency.
Tool details →Vector database
Chroma
Open-source embedding database commonly used for quick RAG prototypes.
Best for: Developers building their first RAG prototype or fast document retrieval experiment who need a vector store running in under 10 lines of Python with no infrastructure setup.
Tool details →Vector database
Weaviate
Open-source vector database and AI-native search platform with hybrid search features.
Best for: Teams building production retrieval systems that need hybrid search — combining dense vector ANN similarity with BM25 keyword search — and want a schema-driven data model with modular extensibility for reranking, generative search, and multi-modal embeddings.
Tool details →Vector database
Milvus
Open-source vector database designed for large-scale similarity search.
Best for: Engineering teams building high-throughput vector search systems at the tens-of-millions to billions-of-vectors scale, where index sharding, distributed query routing, and disaggregated storage and computation are architectural requirements.
Tool details →Vector database
LanceDB
Developer-friendly vector database built around Lance columnar data format workflows.
Best for: AI engineers and data scientists who want vectors, embeddings, and rich metadata to live alongside structured data in a single dataset without running a separate database service.
Tool details →Vector database
pgvector
PostgreSQL extension for storing embeddings and running vector similarity search.
Best for: Teams that already operate a PostgreSQL database and want to add vector similarity search for RAG — keeping embeddings, metadata, and application data in one SQL database without adding a separate vector store service to the infrastructure stack.
Tool details →Inference serving
vLLM
High-throughput open-source LLM serving engine for production and research workloads.
Best for: ML engineers and infrastructure teams deploying open-weight models in GPU server environments where throughput, latency under concurrent load, and GPU memory efficiency are production requirements.
Tool details →Inference serving
SGLang
Fast serving framework and programming interface for language model applications.
Best for: Engineers serving large MoE models (DeepSeek, Mixtral) or applications with high prompt-reuse patterns where RadixAttention prefix caching delivers measurable throughput gains over standard KV cache approaches.
Tool details →Inference serving
Text Generation Inference
Hugging Face server for deploying and serving text generation models.
Best for: Teams serving open-weight Hugging Face models at production throughput — using continuous batching, tensor parallelism, and flash attention — where native Hugging Face Hub integration and the Inference Endpoints managed service are part of the deployment strategy.
Tool details →Inference serving
LiteLLM
Proxy and SDK for routing requests across many LLM providers with OpenAI-compatible interfaces.
Best for: Engineering teams managing multi-provider LLM usage in production who need a single gateway for routing, cost control, fallbacks, and audit logging without vendor lock-in to a single model provider.
Tool details →Inference serving
LocalAI
Open-source OpenAI-compatible local inference server for multiple model types.
Best for: Developers who want to replace OpenAI API calls with a self-hosted drop-in endpoint running on their own hardware — including CPU-only servers, edge devices, and air-gapped environments without CUDA GPUs.
Tool details →Inference serving
BentoML
Model serving platform for packaging, deploying, and operating AI services.
Best for: ML engineers and teams who need to package model serving code — including custom pre/post-processing, multi-model pipelines, and hardware configurations — into reproducible deployment artifacts that run consistently from local development to production.
Tool details →Evaluation and observability
Langfuse
Open-source observability, tracing, prompt management, and evaluation platform for LLM apps.
Best for: Teams shipping LLM applications who need production observability — tracing every generation call, prompt version, retrieval step, and quality score — in a platform that also manages prompt templates and runs automated evaluation pipelines.
Tool details →Evaluation and observability
Phoenix
Arize Phoenix is a source-available observability and evaluation tool for LLM and ML systems.
Best for: Teams who need a source-available LLM observability platform with a built-in web UI for trace inspection, RAG retrieval evaluation, embedding visualization, and experiment comparison — running locally without requiring a cloud account.
Tool details →Evaluation and observability
OpenTelemetry
Open standard and ecosystem for traces, metrics, and logs increasingly used in LLM app observability.
Best for: Engineering teams that want LLM application traces, metrics, and logs to flow into existing observability infrastructure — Datadog, Grafana, Jaeger, Honeycomb, or any OTLP-compatible backend — using the same instrumentation standard as the rest of the service stack.
Tool details →Evaluation and observability
Ragas
Open-source framework for evaluating RAG pipelines and LLM application quality.
Best for: RAG application developers who need an evaluation framework measuring retrieval and generation quality using the RAG triad metrics — context precision, context recall, faithfulness, and answer relevancy — without relying on manual human review of every pipeline output.
Tool details →Evaluation and observability
DeepEval
Open-source LLM evaluation framework for unit-testing model outputs and app behavior.
Best for: Python developers who want to write repeatable unit tests for LLM application behavior — covering answer relevance, hallucination detection, RAG faithfulness, and agent tool-call correctness — using a familiar pytest-style test pattern.
Tool details →Image workflows
ComfyUI
Node-based interface for building advanced local image generation workflows.
Best for: Creators and builders who need precise, reusable image generation pipelines.
Tool details →AI infrastructure
GPUStack
Open-source GPU cluster manager for deploying AI models across owned hardware.
Best for: Builders who want more structure than a single-machine local runtime for self-hosted model serving.
Tool details →MCP Servers
Codebase Memory MCP
High-performance Model Context Protocol server delivering persistent hierarchical memory mapping and contextual intelligence layers directly across local codebases.
Best for: Developers building multi-agent autonomous coding loops, terminal automation layers, or persistent IDE spaces that require a persistent, structured, cross-session memory model of local software codebases.
Tool details →Local benchmarking
whichllm
CLI tool that finds the local LLM that best fits a specific hardware setup, ranked by real recency-aware benchmark results rather than marketing claims.
Best for: Developers and builders who want a hardware-aware recommendation of which local model to run, based on actual benchmark performance on their GPU and RAM configuration.
Tool details →Research agent
Local Deep Research
Open-source iterative deep research agent that runs on local or cloud LLMs -- achieves near-frontier research quality using Ollama, llama.cpp, or commercial APIs with 10+ search backend integrations.
Best for: Developers and researchers who want GPT-Researcher-class deep research capability running against local models (Ollama, llama.cpp) or any OpenAI-compatible endpoint -- offline, private, and without per-query API costs.
Tool details →RAG and knowledge base
MaxKB
Open-source enterprise knowledge base and RAG platform with a visual workflow builder, Ollama integration, MCP server support, and multi-model backend routing for building production AI agent applications.
Best for: Teams building enterprise-grade RAG and knowledge base applications with a visual interface -- document ingestion, multi-model routing, agent workflows, and MCP server integration -- without writing infrastructure code from scratch.
Tool details →Terminal AI agent
Gemini CLI
Open-source terminal AI agent from Google that brings Gemini directly into your command line with MCP server support, multimodal input, agentic task execution, and a 1M-token context window — free to use with a Google account.
Best for: Developers who want a powerful terminal-native AI agent with MCP tool integration, large context windows, and multimodal input (images, PDFs, video) without a paid API subscription.
Tool details →MCP server
Context7
MCP server that pulls live, version-specific library documentation directly into your AI coding agent — eliminating hallucinated APIs by giving Claude, Cursor, Copilot, or any MCP-compatible editor access to up-to-date code docs at query time.
Best for: Developers using AI coding agents (Claude, Cursor, Copilot, Windsurf) who are tired of the agent hallucinating deprecated APIs or outdated method signatures from its training data.
Tool details →MCP server
GitHub MCP Server
GitHub's official Model Context Protocol server that gives AI coding agents authenticated access to repositories, issues, pull requests, code search, and Actions — enabling agents to read, write, and manage GitHub resources directly from any MCP-compatible editor.
Best for: Developers using AI coding agents (Claude, Cursor, Cline, Gemini CLI) who want the agent to read issues, open PRs, search code, and manage GitHub resources without manual copy-paste between the editor and browser.
Tool details →Research agent
GPT Researcher
Open-source autonomous research agent that conducts multi-step deep research on any topic using any LLM — including local Ollama models — with parallel web search, source verification, and structured report generation.
Best for: Developers and researchers who need automated deep research reports with cited sources, supporting both cloud LLMs and local Ollama models for fully private research workflows.
Tool details →Server management
1Panel
Open-source VPS and server control panel with native AI agent support — run Ollama models, manage Docker containers, deploy apps, and configure servers through a modern web UI with built-in local LLM integration.
Best for: Self-hosters and small teams who want a modern server control panel that natively manages Ollama model deployment alongside traditional server infrastructure like web servers, databases, and SSL certificates.
Tool details →Agent framework
Open Multi-Agent
TypeScript multi-agent orchestration framework where a coordinator LLM decomposes a natural-language goal into a task DAG and distributes subtasks to specialized sub-agents — all runnable against local Ollama models or cloud providers.
Best for: Developers who want to build goal-directed multi-agent pipelines in TypeScript without writing orchestration boilerplate, and who need local-LLM support from the start rather than as an afterthought.
Tool details →MCP server
Chrome DevTools MCP
Official MCP server from the Chrome DevTools team that exposes browser debugging capabilities — DOM inspection, console access, network monitoring, and performance profiling — directly to AI coding agents via the Model Context Protocol.
Best for: AI coding agent workflows (Claude Code, Cline, Cursor) where the agent needs to inspect live browser state, debug console errors, or analyze network requests without copy-pasting DevTools output manually.
Tool details →AI notebook
Surf
Local-first personal AI notebook that ingests files and webpages, generates structured notes from them using your choice of open model, and keeps all data on your device — open source, open data, no vendor lock-in.
Best for: Researchers, writers, and knowledge workers who want a private AI note-taking assistant that works with local models like Gemma and DeepSeek instead of sending documents to a cloud API.
Tool details →Healthcare AI
OpenMed
Local-first healthcare AI platform for clinical NLP tasks — named entity recognition, HIPAA PII de-identification, and medical document processing — running 100% on-device with access to 1,000+ medical-domain models.
Best for: Healthcare developers, clinical data engineers, and research teams who need HIPAA-compliant AI processing of medical documents without sending patient data to a cloud API.
Tool details →Agent framework
nanobot
Self-hosted personal AI agent framework in Python — a terminal agent and a local web UI, with built-in tools, persistent memory, scheduled automations, and first-class Ollama support so the whole loop can run on your own hardware.
Best for: Developers who want a persistent personal agent — one that remembers, runs on a schedule, and can execute tools — without renting it from a cloud service or wiring the orchestration themselves.
Tool details →Chat workspace
Chatbox
Cross-platform desktop AI client that talks to local runtimes and cloud providers through one interface, with a built-in Ollama provider, document knowledge bases, and local storage of conversations.
Best for: People who want a polished desktop chat application over their local models — a single window for Ollama, LM Studio and cloud providers, without running a web server or living in a browser tab.
Tool details →