Data & privacy

What every message records

Tokens in, cached tokens, tokens out, the model used, latency and the exact cost in USD. Without those numbers your margin is a theory.

tokens_in     8,412
cached_tokens 8,398   (99.8%)
tokens_out      146
model         qwen3.7-flash
cost_usd    $0.000061
latency        1,204 ms

How visitors are identified

By a salted hash of IP address and user agent, truncated. Never the raw address. Pseudonymisation is an Article 32 measure and it is what lets rate limiting survive a deletion request.

Retention

Set a window and conversations delete themselves when they reach it. Until then they are yours to read, export or remove by hand at any time.

Where it lives

EU infrastructure, so data stays in the EU by default. Where a model provider processes outside the EU, that transfer is covered and named on the sub-processor list.

Export and delete

Any conversation, any time, from the dashboard. Close your account and conversation data goes with it.