Last updated 07-22-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
Twinny
Twinny is a free, open-source Visual Studio Code extension for local and API-hosted code completion and chat. It targets developers who want Copilot-style assistance without a paid subscription, using inference providers you control such as Ollama or cloud APIs from OpenAI, Anthropic, and others.
The extension supports fill-in-the-middle completions, a sidebar chat for explanations and refactors, workspace embeddings for context-aware suggestions, and optional Symmetry network integration for peer-to-peer inference sharing. It follows the OpenAI API standard, so you can point it at localhost endpoints or hosted model providers.
Twinny has been archived. The GitHub repository was archived on November 24, 2025, and twinny.dev now displays a shutdown notice. The VS Code marketplace listing remains available, but the project is no longer actively developed.
Fill-in-the-middle completions as you type, single-line or multiline
Sidebar chat for explanations, tests, refactors, and commit messages
Works with Ollama, OpenAI, Anthropic, OpenRouter, Mistral, Groq, and more
Workspace embeddings pull project context into every suggestion
Symmetry network hooks let you share or borrow P2P inference capacity
Runs online or offline when paired with a local inference provider
Completely free and MIT licensed with no subscription required for the extension itself.
Supports many inference backends, including fully local setups through Ollama.
Fill-in-middle completions and sidebar chat cover both inline and conversational coding workflows.
The project has been archived and no longer receives active development or support.
You must configure and run your own inference provider; Twinny does not include hosted models.
The VS Code marketplace listing still describes the project as actively developed, which conflicts with the archive status.
Is Twinny free to use?
Yes. Twinny is a free, open-source Visual Studio Code extension listed on the VS Code Marketplace. You still need your own inference provider, such as Ollama locally or a paid API from OpenAI or Anthropic, to run the models.
What inference providers does Twinny support?
Twinny supports a localhost OpenAI/Ollama-compatible API by default, plus OpenAI, Anthropic, OpenRouter, Deepseek, Cohere, Mistral AI, Perplexity, and Groq. You configure the provider, model, port, and path in the extension settings.
How do I install Twinny in VS Code?
Install Twinny from the Visual Studio Code Marketplace or Open VSX for VSCodium. The docs recommend pairing it with Ollama: pull a model like codellama:7b-instruct, run it locally, then open the twinny sidebar and look for the robot icon in the status bar.
Does Twinny work offline?
Yes, when you use a local inference provider such as Ollama or another OpenAI-compatible server on your machine. Cloud API providers require an internet connection and their own API keys.
Is Twinny still being developed?
No. Twinny has been archived. The GitHub repository was archived on November 24, 2025, and twinny.dev displays a shutdown notice. The marketplace listing remains installable, but the maintainers are no longer shipping updates.
What is the Symmetry network in Twinny?
Symmetry is a decentralized peer-to-peer network Twinny integrated for sharing AI inference resources. Twinny docs describe it as a way to connect nodes, share compute, and route inference across the network when configured with the Symmetry client.