Terms of service

Last updated: 2026-10-24 (draft)

These terms cover use of the inference service run by Cold-Serve Labs ("Cold Serve", "we").

The service is provided as is. We aim for high uptime but do not guarantee it.
You are responsible for your prompts and for how you use outputs. Model outputs can be wrong.
Do not use the service for illegal content, or for attacks on us or others. Each model's own license and use policy also applies.
Prices are listed per model, per million tokens. Usage is billed from token counts.
We may change these terms and will post the date of the change here.
These terms are governed by the laws of Japan.