Skip to main content
Deepshi is built for people who want AI without surveillance. The principle is simple: the best way to keep your data private is not to collect it in the first place.

​
What we don’t do

No training on your data

Your prompts and completions are never used to train models, Deepshi’s or anyone else’s.

No content filtering

Deepshi’s own models answer without the moderation and refusals you hit on mainstream APIs.

​
How requests are handled

  • Requests travel over HTTPS/TLS and are processed for inference only.
  • Deepshi’s own models run on infrastructure Deepshi controls, so your prompts aren’t handed to third parties to fulfill them.
  • Operational metadata (authentication, usage metering, billing, and abuse prevention) is processed to run the service. This is separate from prompt and completion content.
When you call a third-party model (for example a frontier model from another provider) through the API, that request is fulfilled by the upstream provider and is subject to their handling. Use Deepshi’s own models (Deepshi 2.0, Deepshi 3.0) for the strongest privacy posture.

​
Logging

By default, Deepshi does not retain the content of your prompts and responses. Usage records (tokens, cost, latency, model, and status) are kept so you can see your spend and history in the dashboard.

​
Learn more

See deepshi.ai for Deepshi’s full privacy stance, including third-party verification of its privacy claims.