
Tabby
TabbyML's open-source, self-hosted AI coding assistant with code completion, chat and IDE extensions.
Tabby is TabbyML's open-source, self-hosted AI coding assistant: you run its code completion and chat server and connect VS Code, JetBrains IDEs and other editors to it. The Community license is free for up to 5 users; Team is $19 per seat a month for up to 50 users, and Enterprise is custom-priced and billed annually.
About Tabby
TabbyML's docs describe Tabby as a server each team sets up for LLM-powered code completion. It suggests multi-line code and full functions as you type, answers general and codebase questions in chat and, through its Answer Engine, inside the IDE, edits code on request with an inline editing command, and uses Context Providers to pull in documentation, configuration files and other sources.
The docs cover installing with Docker, Homebrew, Hugging Face Space and other methods. Linux and Windows get standalone builds in CPU-only and CUDA versions, Linux adds a Vulkan build for non-NVIDIA GPUs, and on Apple M1/M2 it installs through Homebrew. Once running, the server is reached by default at localhost:8080, and editors connect with its URL and an account token.
The homepage names the editors it supports: VS Code, Neovim, IntelliJ, Eclipse, Android Studio, AppCode, CLion, GoLand, PhpStorm, PyCharm, Rider, RubyMine and WebStorm. The VS Code extension is on the Visual Studio Marketplace and Open VSX, and the IntelliJ Platform plugin, from JetBrains Marketplace, works with IDE builds from 2023.1. The docs say Tabby works with major coding LLMs such as CodeLlama, StarCoder and CodeGen and keep a curated directory of recommended models.
Paid tiers are licenses for the same server: the Community license is free for anyone, while Team and Enterprise licenses are paid, add features and are uploaded on the server's Subscription page. Team is listed at $19 per seat a month, with no yearly price shown, for up to 50 users with email support. Enterprise is custom-priced and billed annually, with unlimited users and a dedicated Slack channel for support.
Sources: tabbyml.com, /pricing, tabby.tabbyml.com/docs/welcome, tabby.tabbyml.com/docs/quick-start/installation/apple, tabby.tabbyml.com/docs/quick-start/installation/linux, tabby.tabbyml.com/docs/quick-start/installation/windows, tabby.tabbyml.com/docs/extensions/installation/vscode, tabby.tabbyml.com/docs/extensions/installation/intellij and tabby.tabbyml.com/docs/administration/license, read 2026-10-27
Tabby pros and cons
Pros
- The free Community plan includes code completion, the Answer Engine, inline chat and Context Providers.
- The homepage says Tabby needs no external database or cloud service.
- CPU-only builds for Linux and Windows run without a GPU, and GPU setups can use CUDA, Vulkan, Metal or ROCm.
- A paid license can be reset back to the free Community license with one button.
Cons
- The Community plan is capped at 5 users and Team at 50.
- SSO, an authentication domain, IDE telemetry policy enforcement and roadmap prioritization are only in Enterprise.
- On Apple M1/M2, the docs expect enough compute only for individual use and recommend Docker hosting with CUDA or ROCm for a team.
- The IntelliJ Platform plugin requires installing Node.js 18.0 or higher.
Tagged
- Self-hostable
- Supports Local Models
- VS Code Extension
- JetBrains Plugin
- Open Source
Tabby alternatives
The closest alternatives to Tabby.
Related reading
Featured on Vibedonalds
Own Tabby? Add this badge to your site to show you’re listed — and link back to your profile here.
<a href="https://vibedonalds.com/tools/tabby" target="_blank" rel="noopener">
<img src="https://vibedonalds.com/badge/featured-on-vibedonalds.svg" alt="Tabby — Featured on Vibedonalds" width="240" height="60" loading="lazy" />
</a>Frequently asked questions
- What is Tabby?
- Tabby — TabbyML's open-source, self-hosted AI coding assistant with code completion, chat and IDE extensions.
- What platforms does Tabby support?
- Tabby's own site lists it for macOS, Windows, Linux, VS Code extension and JetBrains plugin.
- What are the alternatives to Tabby?
- The closest alternatives to Tabby are Continue, GitHub Copilot.
- What are the downsides of Tabby?
- The Community plan is capped at 5 users and Team at 50. SSO, an authentication domain, IDE telemetry policy enforcement and roadmap prioritization are only in Enterprise. On Apple M1/M2, the docs expect enough compute only for individual use and recommend Docker hosting with CUDA or ROCm for a team. The IntelliJ Platform plugin requires installing Node.js 18.0 or higher.
- What is Tabby AI?
- Tabby is TabbyML's open-source, self-hosted AI coding assistant. You run its server, which provides code completion, chat and an Answer Engine, and connect your editor through an extension for VS Code, the IntelliJ Platform IDEs or another supported editor.
- How do I download and install Tabby?
- The docs install Tabby with Docker, with Homebrew on Apple M1/M2 (brew install tabbyml/tabby/tabby), on Hugging Face Space and in other ways, or from standalone Linux and Windows builds on the GitHub release page. Once it runs, the server is at localhost:8080 by default, and editors connect with its URL and an account token.
- Is Tabby free?
- The open-source server is free under the Community license, which anyone can use, for up to 5 users. Team is listed at $19 per seat a month for up to 50 users, and the pricing page shows no yearly price for it. Enterprise is custom-priced and billed annually.
- Which IDEs does Tabby support?
- TabbyML's homepage names VS Code, Neovim, IntelliJ, Eclipse, Android Studio, AppCode, CLion, GoLand, PhpStorm, PyCharm, Rider, RubyMine and WebStorm among the IDEs it supports. The VS Code extension is on the Visual Studio Marketplace and Open VSX; the IntelliJ Platform plugin comes from JetBrains Marketplace and needs IDE build 2023.1 or later.
- Does Tabby need a GPU?
- Not necessarily: Linux and Windows have CPU-only builds next to CUDA builds, and Linux has a Vulkan build for non-NVIDIA GPUs. The docs say Apple M1/M2 compute is likely enough only for individual use and recommend Docker hosting with CUDA or ROCm for a team.