Twinny

twinny’s editor extension can provide code completion and answer questions based on the code context, as well as display modification and review results. It currently supports Ollama, llama.cpp, LM Studio, and OpenAI-compatible services, while the team gateway can also organize internal computing resources and backends.

Extensions and gateway operations can be managed by the team, but whether data leaves the network depends on the specifically configured model services; not all cloud‑compatible endpoints can be considered locally offline. Features such as team policies, model locking, audit logs, and plugins are available based on the license; generated code still requires review and testing.

Developers and self-hosted AI coding teams

2026-10-25 The official website lists MIT‑extended free access; twinny-server offers up to 5 developers per gateway at no cost. Team provides a 30‑day trial with 25 seats, with no card required; upon expiration, it reverts to the free tier. The pricing for page seats and subscription periods varies, and a fixed unit price has not been uniformly applied—please confirm according to the license settlement page. Model resource costs are billed separately.