# Tabby Self-hosted AI coding assistant with code completion, chat, and agent capabilities that runs entirely on your infrastructure. ## Agent Decision Summary - Risk level: moderate - Source confidence: high - Recommended workflows: Coding agent workflow, Local or private AI stack - Permission surface: shell/files - Agent JSON: https://www.openagent.bot/agents/tabby.agent.json ## Summary Tabby is an open-source, self-hosted AI coding assistant that provides code completion, chat, and agent capabilities without any external dependencies or data leaving your infrastructure. It supports VS Code, JetBrains, Vim, Emacs, and other editors, and works with self-hosted models for fully air-gapped operation. ## Guide Tabby is the leading self-hosted alternative to cloud-based AI coding assistants. For teams that cannot send code to external services due to compliance, security, or infrastructure constraints, Tabby provides a complete AI coding assistant that runs entirely on their own hardware. ### What it is Tabby is an open-source, self-hosted AI coding assistant that provides code completion, chat, and agent capabilities. It runs entirely on your infrastructure with no external dependencies. ### Why it matters Most AI coding tools require sending code to external APIs. Tabby is designed for teams that cannot or will not let code leave their network. Combined with support for any open-source model, Tabby is the most practical option for enterprise, government, and regulated industry deployments. ### How it works Start with one safe workflow for Tabby. Inspect official setup instructions, required credentials, execution logs, approval points, and failure recovery before expanding from a sandbox task into production automation. ## Use Cases - Regulated industry deployment: Deploy Tabby behind your firewall for healthcare, financial services, or government work where data cannot leave controlled infrastructure. - Offline development environments: Run Tabby in fully air-gapped environments with no internet connectivity, using locally hosted open-source models. - Custom model serving: Fine-tune and serve custom models optimized for your codebase, domain-specific languages, or internal frameworks. ## Alternatives - Use Continue for a lighter self-hosted option vs Continue: Tabby provides a full self-hosted backend with completion, chat, and agent capabilities. Continue offers IDE-native agents with optional CI checks. - Use Cline for cloud-connected agent capabilities vs Cline: Tabby is designed for self-hosted, offline-first deployments. Cline offers richer autonomous agent features but requires cloud API access. ### Getting Started - Read the documentation: https://tabby.tabbyml.com/docs - Inspect the repository: https://github.com/TabbyML/tabby ### FAQ - What should I check before using Tabby? - Start with one safe workflow for Tabby. Inspect official setup instructions, required credentials, execution logs, approval points, and failure recovery before expanding from a sandbox task into production automation. - Is Tabby open source? - Tabby is listed with Apache-2.0 based on the official source links in this profile. Re-check the repository, model card, or docs before production use. - What hardware does Tabby need? - Tabby requires a machine with a GPU for good performance. It supports NVIDIA GPUs and Apple Silicon, and can run CPU-only with reduced speed. - Can Tabby work fully offline? - Yes. Tabby is designed for fully offline operation with no internet access required once deployed. ## What It Does Tabby is an open-source, self-hosted AI coding assistant that provides code completion, chat, and agent capabilities. It runs entirely on your infrastructure with no external dependencies. ## How To Evaluate Start with one safe workflow for Tabby. Inspect official setup instructions, required credentials, execution logs, approval points, and failure recovery before expanding from a sandbox task into production automation. ## Why It Matters Tabby is the leading self-hosted alternative to GitHub Copilot and Cursor. For teams with strict data sovereignty requirements, air-gapped environments, or enterprise compliance needs, Tabby provides a complete AI coding assistant that runs entirely on their own infrastructure. ## Best For - Enterprise teams with strict data sovereignty and compliance requirements - Developers working in air-gapped or offline environments - Teams that want full control over their AI coding infrastructure and model selection ## Not For - Individual developers who prefer cloud-hosted coding assistants - Users who want frontier models without managing local GPU infrastructure ## What It Actually Does - Fully self-hosted: Runs entirely on your infrastructure with no external API calls or data leaving your network. - Why it matters: For regulated industries (healthcare, finance, government), Tabby is one of the few viable AI coding assistants because no code ever leaves the organization. - Multi-editor support: Works with VS Code, JetBrains, Vim, Emacs, and other editors through standardized plugins. - Why it matters: Teams with diverse editor preferences can standardize on one self-hosted AI backend. - Code completion + chat + agent: Provides inline code completion, conversational chat, and agentic task execution from a single self-hosted service. - Why it matters: One deployment covers the full spectrum of AI-assisted development without relying on external services. - Open model support: Works with any open-source model including Llama, DeepSeek, Qwen, and fine-tuned custom models. - Why it matters: Teams can choose, fine-tune, or build custom models for their specific codebase and domain. ## Typical Use Cases - Air-gapped development: Deploy Tabby in an air-gapped environment with no internet access, running entirely on local GPU infrastructure. - Enterprise compliance: Meet SOC 2, HIPAA, or GDPR requirements by keeping all code and AI processing within your controlled infrastructure. - Custom model fine-tuning: Fine-tune open models on your codebase for better completion quality, then serve them through Tabby. ## How It Compares - Choose Tabby for self-hosted data sovereignty vs GitHub Copilot: Tabby runs entirely on your infrastructure with no data leaving your network. Copilot sends code snippets to GitHub's servers for processing. ## Fit Matrix - Coding agent workflow: strong. Tabby has multiple signals for coding agent workflow, including matching tags, capabilities, category, or positioning. Required check: Run a small repository change and inspect the diff, tests, and rollback path. - Local or private AI stack: strong. Tabby has multiple signals for local or private ai stack, including matching tags, capabilities, category, or positioning. Required check: Verify hardware requirements, data path, storage, and whether all calls stay in your environment. - Connector or protocol layer: partial. Tabby has at least one signal for connector or protocol layer, but should be checked against a real task before adoption. Required check: Connect one low-risk service, then inspect schemas, auth scope, errors, and logs. - Reusable skill workflow: partial. Tabby has at least one signal for reusable skill workflow, but should be checked against a real task before adoption. Required check: Run one skill end to end and check whether it produces evidence or structured output. - Browser automation: weak. Tabby is not primarily positioned for browser automation in the current metadata. Required check: Run one non-sensitive website task and inspect clicks, waits, retries, and changed URLs. - Evaluation and observability: weak. Tabby is not primarily positioned for evaluation and observability in the current metadata. Required check: Add one repeatable test case and confirm results can run again in review or CI. ## Evidence - verified: Tabby is listed as open source. Source: License metadata: Apache-2.0 - verified: Tabby has a recorded GitHub repository: TabbyML/tabby. Source: Resource facts and GitHub source link. - inferred: Tabby supports these recorded deployment modes: local, self hosted, cloud. Source: OpenAgent decision signal metadata. - inferred: Tabby is tagged with local inference, workflow orchestration capabilities. Source: OpenAgent capability taxonomy. ## Missing Checks - Repository freshness has not been recorded. ## Next Actions - Inspect repository: https://github.com/TabbyML/tabby - Open Homepage: https://tabby.tabbyml.com - Read setup docs: https://tabby.tabbyml.com/docs - Deploy Tabby: docker run -it --gpus all -p 8080:8080 tabbyml/tabby serve --model StarCoder-15B ## Command Line ### Deploy Tabby Deploy Tabby via Docker with GPU support. Once running, install the editor extension and connect to your self-hosted instance. ```bash docker run -it --gpus all -p 8080:8080 tabbyml/tabby serve --model StarCoder-15B ``` ## Facts - Category: agents - Resource type: agent - Open source: yes - License: Apache-2.0 - Last verified: 2026-06-04 - GitHub repo: TabbyML/tabby - GitHub stars: 32200 ## Capabilities - local-inference - workflow-orchestration ## Structured Use Case Tags - local-ai - self-hosted-ai - developer-workflow ## Getting Started - Open the GitHub repository: https://github.com/TabbyML/tabby - Read the documentation: https://tabby.tabbyml.com/docs - Visit the project website: https://tabby.tabbyml.com ## Links - GitHub: https://github.com/TabbyML/tabby - Homepage: https://tabby.tabbyml.com - Docs: https://tabby.tabbyml.com/docs ## Structured Outputs - JSON: https://www.openagent.bot/agents/tabby.json - Markdown: https://www.openagent.bot/agents/tabby.md - Agent JSON: https://www.openagent.bot/agents/tabby.agent.json - Canonical: https://www.openagent.bot/agents/tabby