Terms of Service — nova26 inference endpoints

Effective 2026-10-25. These terms cover the model endpoints operated by nova26 (the "provider") and listed on marketplaces including OpenRouter and RapidAPI.

The service

The API serves one model, named in the listing and in every /v1/models response: Ternary-Bonsai-2-2B, served as nova26/ternary-bonsai-2-2b. The provider may update serving internals; the served model's name changes only with a listing update.

Metering

Calls are metered by the marketplace through which they are routed, per the plan or pricing shown on that marketplace's listing: each request to POST /v1/chat/completions counts as usage at the listed rates. GET /v1/models and GET /health are not metered.

Your content

Callers are responsible for the content they send. The provider's privacy policy governs what the provider does with it: metadata-only logging, no training use, zero content retention. See Privacy & data retention policy.

Availability

The service is provided on a best-efforts basis during its initial listing period; no uptime guarantee is made until the provider publishes one on the listing.

Billing

Billing, refunds, and disputes run through the marketplace that routed the call, under that marketplace's own terms; the provider does not take payment details directly.

Contact

Questions: [email protected]