Token access
Pay for the
tokens you use.
No subscription trap. Buy inference capacity when you need it. During the launch beta, eligible requests are free until the server-side beta window closes.
Starter · 1M TOKENS
$5one-time
1M inference tokens
OpenAI-compatible API
Streaming completions
Vision-ready mmproj path
No prompt/response retention
Builder · 5M TOKENS
$20one-time
5M inference tokens
OpenAI-compatible API
Streaming completions
Vision-ready mmproj path
No prompt/response retention
Scale · 25M TOKENS
$75one-time
25M inference tokens
OpenAI-compatible API
Streaming completions
Vision-ready mmproj path
No prompt/response retention
Integer accounting
1 credit = 1,000 reported tokens. Usage is settled from the model server’s usage block.
No surprise charges
Packages are prepaid. Requests fail before dispatch when the paid balance cannot cover the reservation.
Stateless by design
Payment/account metadata and bounded usage numbers are retained; prompt and response content is not.