Model / Provider Disclosure

Effective date: May 11, 2026

Current provider

NoLog routes inference through a backend proxy to a zero-retention provider. The current production provider is configured server-side and may change without an app update.

Current model family

NoLog is currently configured for DeepSeek V4 Flash through CanopyWave for text-only chat. Image and PDF-rendered-image requests may be routed to Kimi K2.6 through CanopyWave. Instant mode disables thinking. Thinking mode requests provider reasoning output when supported.

Thinking output

When Thinking mode returns reasoning text, NoLog shows it in a collapsed thinking block and stores it only with the local chat on this device. NoLog does not store thinking output on the backend.

AI limitations

AI output can be wrong, incomplete, or misleading. Always verify important information independently.