Ollama
Run open LLMs locally with a single command.

Open-source, self-hosted AI coding assistant you run on your own hardware
Tabby is a strong choice for teams that need a genuinely open-source (Apache 2.0), self-hosted coding assistant where code never leaves their infrastructure, and it runs on modest consumer GPUs with broad IDE support. The trade-offs are real: it requires GPU and DevOps effort to run and maintain, completion quality depends on the open model you choose (generally below frontier proprietary tools), and it focuses on completion and chat rather than acting as an autonomous agent.
Tabby is an open-source, self-hosted AI coding assistant built in Rust that runs entirely on your own hardware with no cloud dependency, offering code completion, inline chat, and a shareable Answer Engine over your codebase. It is Apache 2.0 licensed, model-agnostic, runs on consumer NVIDIA or Apple Silicon GPUs, and supports 12-plus IDEs. It is ideal for privacy-conscious and regulated teams wanting a Copilot alternative, though it requires GPU and DevOps effort.
Tabby (by TabbyML) is an open-source, self-hosted AI coding assistant positioned as a privacy-first alternative to GitHub Copilot. Written in Rust and fully self-contained, it runs on your own infrastructure with no external database or cloud dependency, so code never leaves your environment. Its Apache 2.0-licensed core provides real-time inline code completion, inline chat tied to your code, and an Answer Engine that offers a shareable natural-language Q&A layer over your indexed repositories. Tabby is model-agnostic and runs on consumer-grade GPUs, supporting both NVIDIA (CUDA) and Apple Silicon (Metal) without a GPU cluster. It works with open code LLMs such as CodeLlama, StarCoder, Qwen, and DeepSeek, indexes Git repositories (including multi-branch indexing in recent releases), and pulls in additional context from docs, config files, external APIs, and GitLab merge requests, issues, and commits. It ships plugins for 12-plus IDEs, including VS Code, JetBrains, and Vim/Neovim, and includes SSO/OAuth, team administration, and analytics in the free self-hosted build. The project is actively developed, with tens of thousands of GitHub stars and a large contributor base. The self-hosted community edition is free and unlimited, a managed cloud Team plan runs around $24 per user per month, and enterprise pricing is custom. Tabby is ideal for privacy-conscious and regulated teams that want a Copilot alternative on their own GPUs, but it requires GPU and DevOps effort to run, and completion quality depends on the underlying open model rather than a frontier proprietary one.
Tabby is an open-source, self-hosted AI coding assistant built in Rust that runs entirely on your own hardware, offering code completion, inline chat, and a shareable Answer Engine over your codebase. It is Apache 2.0 licensed, model-agnostic, runs on consumer NVIDIA or Apple Silicon GPUs, and supports 12-plus IDEs. The self-hosted edition is free and unlimited; a managed cloud plan runs about $24 per user per month. It is ideal for privacy-first teams willing to run their own infrastructure.
Tabby is developed by TabbyML Inc., founded by Meng Zhang and Lucy Gao, as an open-source, self-hosted alternative to cloud coding assistants.
The project is actively developed with tens of thousands of GitHub stars and a large contributor community, and the company also offers managed cloud and enterprise editions.
Tabby provides real-time inline code completion, inline chat tied to your code, and an Answer Engine that offers shareable natural-language Q&A over indexed repositories. It is model-agnostic, works with open LLMs like CodeLlama and StarCoder, and runs on consumer NVIDIA or Apple Silicon GPUs.
It indexes Git repos (including multi-branch), ingests docs and context providers, ships plugins for 12-plus IDEs, and includes SSO/OAuth, team admin, and analytics in the free self-hosted build. The core is Apache 2.0; some enterprise code is under a separate commercial license.
Privacy-sensitive and regulated teams, enterprises, and self-hosting developers who want a Copilot alternative without sending code to third-party clouds.
Developers on privacy-conscious or regulated teams using self-hosted completions and chat in their IDE.
Engineering leaders and platform or security teams choosing a self-hostable coding assistant.
Open-source advocates, security and compliance stakeholders, and DevOps engineers.
A privacy-sensitive or regulated engineering organization with GPU and DevOps resources that wants an open-source, self-hosted Copilot alternative keeping all code in-house.
TabbyML raised a $3.2 million seed round in October 2023, led by Yunqi Partners with ZooCap. Some aggregators cite a larger total around $7.2 million across rounds, but only the $3.2 million seed is clearly documented.
Yes. The self-hosted community edition is free under Apache 2.0, with unlimited users and completions, including codebase indexing, SSO, and team administration. Managed cloud and enterprise plans are paid.
Tabby runs on consumer-grade GPUs, supporting both NVIDIA (CUDA) and Apple Silicon (Metal), so no GPU cluster is required, but you do need a suitable GPU and some DevOps effort.
It offers plugins for 12-plus IDEs, including VS Code, JetBrains, and Vim/Neovim.
No. Tabby is fully self-contained and runs on your own infrastructure with no external database or cloud dependency, so code stays in-house, which is its core privacy advantage.
Not primarily. Tabby focuses on code completion, inline chat, and an Answer Engine over your codebase rather than autonomously shipping full pull requests.
Side-by-side pages for pricing, features, and best-fit use cases.
Run open LLMs locally with a single command.
AI coding agent built for JetBrains IDEs
High-throughput open-source LLM inference engine
Blazing-fast, open-source code editor with built-in AI agents and real-time collaboration