Subprocessors
Last updated:
These are the third parties that help us run PerToken. We never provide your content to any of them for model training.
Current subprocessors
| Provider | Purpose | Data processed | Location |
|---|---|---|---|
| Cloudflare, Inc. | API gateway (Workers), request routing, usage metering (Durable Objects, Analytics Engine), batch job storage, platform logs, encrypted tunnels to GPU servers, DNS, and website hosting | API content in transit, batch job data, usage metadata, API key hashes, network metadata | Global network |
| Supabase, Inc. | Account authentication and database for account and usage records | Name, email address, sign-in records, hourly usage rollups | [SUPABASE PROJECT REGION] |
| Vast.ai and independent GPU hosts on its marketplace | Rented GPU servers that run the models | API content, processed in memory while a request runs | Varies by host |
| Functional Software, Inc. (Sentry), when enabled | Error monitoring for the API gateway | Scrubbed error reports without request bodies, credentials, cookies, or query strings | [SENTRY DATA REGION] |
| [PAYMENT PROCESSOR] | Payment processing | Billing contact and payment details | [PAYMENT PROCESSOR] |
| [TRANSACTIONAL EMAIL PROVIDER] | Sign-in and account emails | Email address and message content we send | [TRANSACTIONAL EMAIL PROVIDER] |
Other providers that do not receive customer data
- Hugging Face, Inc.: source of the open model weights our GPU servers download. No customer data is sent to Hugging Face.
Marketplaces and routers
OpenRouter, Inc. is not our subprocessor. If you reach PerToken through OpenRouter, OpenRouter is the party sending requests to us, and it receives the responses and usage data for those requests. OpenRouter's own terms and privacy policy govern its handling of your data.
Changes
We update this page before a new subprocessor starts processing customer data. To ask about a subprocessor, email support@pertoken.ai.