Tabnine’s code completion AI uses both cloud and local models. For both models, your team’s code and AI training data are only used to provide the service, and your code is never shared nor used as part of Tabnine’s open-source trained AI.
When using Tabnine Cloud, local context from your file is sent to compute completion predictions.
The context is built from the currently opened file, including related and imported type declarations (i.e., function/objects/classes declarations).
In both models, local context is not stored on Tabnine servers, not even in diagnostics logs.
If sending code to a cloud service is not possible, we also offer a self-hosted option.
Both models are enabled by default, but you can opt out of either model (or both) at any time.