SUB-PROCESSORS
LAST UPDATED: 11 OCTOBER 2026
This page is the authoritative list referred to by the Privacy Policy (Section 6) and the Data Processing Agreement (Section 6 and Annex B). We give organizations 30 days' notice by email before adding or replacing a sub-processor, and update the date above whenever this page changes.
1. OUR SUB-PROCESSORS
Providers to which Vibetron entrusts data in order to run the Service.
| Sub-processor | Purpose | Location | Transfer tool | Since |
|---|---|---|---|---|
| Hetzner Online GmbH | Server hosting and storage (Falkenstein, Germany) | Germany (EU) | Not needed (EU) | Launch |
| Cloudflare, Inc. | DNS, reverse proxy, DDoS protection and WAF for api.vibetron.ai | USA (global network) | EU-US Data Privacy Framework; Standard Contractual Clauses | Launch |
| Resend, Inc. | Transactional email (login codes, budget and operational notices); never carries the content of your trons | USA | Standard Contractual Clauses | Launch |
Removed: Google Fonts (web fonts loaded from Google's servers, which received the IP address of visitors) — the fonts are served from our own server since October 2026.
2. AI MODEL PROVIDERS YOU CAN CONNECT
These are not our sub-processors: you choose them, you hold the account and the key, and your contract with them governs what they do with your data. Vibetron transmits your agents' calls to them from our server in Germany, on your instruction, and never with a key of its own. The table is generated from the Service's own configuration, so it is always the list an agent can actually be pointed at. The location is the country from which the endpoint is operated; by choosing an endpoint you choose where that agent's data goes. Endpoints operated from mainland China are shown as «China (mainland)».
Loading the list of AI model providers…
3. OTHER SERVICES YOU CONNECT
Also not our sub-processors, and used only if you configure them: Telegram and Discord (through the bot account you created with them), Model Context Protocol (MCP) servers and plugins that call external services, and local models you run yourself (called only at a publicly reachable address).
4. SERVICES THAT RECEIVE NO PERSONAL DATA
To keep its price list current, the Service reads public model catalogues (models.dev, and the LiteLLM price file on GitHub). These requests carry no personal data.