Open Source

Tabby

Self-hosted AI code completion server

Is Tabby open source?

Yes. Tabby is open source software released under the Apache-2.0 license and developed by TabbyML. The ee/ enterprise directory is under a separate license. The source code is public in its official repository.

Advertisement

Overview

Tabby is an open source AI coding assistant developed by TabbyML and released on GitHub under the Apache-2.0 license, with roughly 33,000 stars as of mid-2026. It’s written in Rust and distributed as a single self-contained server binary or Docker image that bundles model serving, an inference runtime, and a web dashboard, with no external database or cloud service required to run it. Code outside of the repository’s ee/ (enterprise edition) directory is Apache-2.0 licensed; that directory carries its own separate license covering additional team-management features, and bundled third-party components retain their original licenses.

Where Tabby differs sharply from other tools already covered on this site is scope: Aider, Cline, Continue, and OpenHands are agentic editors that plan and execute multi-file changes across a codebase, often driving a remote frontier model through an API. Tabby instead focuses specifically on the inline, ghost-text autocomplete experience that GitHub Copilot popularized — predicting the next few lines as you type — and is built to run the completion model itself on hardware you own, down to a single consumer-grade GPU, rather than sending completion requests to a vendor’s cloud. It ships official IDE extensions for VS Code, JetBrains, and Vim/Neovim that call its local API in real time, plus a chat panel and a retrieval layer (“Answer Engine”) that can index a team’s repositories and documentation to ground suggestions and answers in internal code.

Key Features

  • Self-hosted inference — Runs the full completion pipeline on your own server or workstation, including consumer GPUs, with no mandatory cloud dependency
  • IDE extensions — Official plugins for VS Code, JetBrains IDEs, and Vim/Neovim
  • Inline code completion — Real-time, context-aware ghost-text suggestions similar to Copilot’s core experience
  • In-IDE chat — Ask questions and get explanations without leaving the editor
  • Repository-aware context — Indexes connected repositories and docs so completions reflect internal codebases, not just public training data
  • Model flexibility — Works with a range of open-weight and hosted code models, swappable via configuration
  • No external dependencies — Runs without an external DBMS or third-party cloud service, simplifying air-gapped or on-premises deployment
  • Team dashboard — Web UI for managing users, monitoring usage, and configuring models (enterprise features live under a separate ee/ license)

Use Cases

  • Air-gapped or regulated environments — Teams that cannot send source code to a third-party API but still want Copilot-style completions
  • GPU-owning teams — Organizations with existing on-prem or cloud GPU capacity looking to avoid per-seat completion subscriptions
  • Completion-focused workflows — Developers who want fast inline suggestions rather than a full agentic multi-file editing loop
  • Internal codebase grounding — Teams wanting completions and chat answers informed by their own private repositories rather than only public code

Who It’s For

Tabby is for organizations and individual developers who want the specific experience of GitHub Copilot’s inline autocomplete — not an agentic assistant that edits files on its own — but with the model running on infrastructure they control instead of a vendor’s servers. It’s a strong fit for regulated industries, security-conscious teams, or anyone with spare GPU capacity who wants completion latency and data locality under their own control. Developers who want an agent that plans and executes multi-file changes should look instead at Aider, Cline, or OpenHands, which Tabby is intentionally narrower than by design.

Advertisement

This is a free alternative to: GitHub Copilot, Cursor

Similar software: Development Tools

Aider

AI pair programming in your terminal

Aider is a free, open source AI pair programming tool for the terminal. It works with Claude, GPT, Gemini, DeepSeek, and local LLMs to edit code in git repos.

Open Source Apache-2.0

Cline

Autonomous coding agent as SDK, IDE extension, or CLI

Cline is an open source autonomous coding agent for VS Code, JetBrains, CLI, or SDK use. Supports Claude, GPT, Gemini, and local models.

Open Source Apache-2.0

Continue

Open source coding agent for VS Code and JetBrains

Continue is an open source AI coding agent for VS Code and JetBrains with autocomplete, chat, and agentic editing. Unmaintained since mid-2026.

Open Source Apache-2.0

goose

Open source, general-purpose AI agent with a desktop app and CLI

goose is a free, Apache-2.0 AI agent built at Block and now part of the Linux Foundation's Agentic AI Foundation, with a desktop app, CLI, and MCP extensions.

Open Source Apache-2.0

OpenCode

Open source AI coding agent for the terminal, desktop, and IDE

OpenCode is a free, MIT-licensed AI coding agent for the terminal, desktop, and IDEs that works with 75+ model providers, local models, and existing plans.

Open Source MIT

OpenHands

AI-driven autonomous development assistant

OpenHands is an open source AI coding agent (formerly OpenDevin) that writes code and runs sandboxed commands, self-hostable via Docker with a browser UI.

Open Source MIT