These are the third parties that process data on our behalf to deliver the Services, as referenced in our Privacy Policy. We maintain this page rather than listing providers inside the policy itself, so that it stays accurate. Locations given are the provider's principal processing jurisdiction; providers may use their own subprocessors and additional regions.
1. Infrastructure and platform
| Provider | Purpose | Data | Location |
|---|---|---|---|
| Supabase | Authentication and primary database | Account data, prompts, chat history, project metadata | United States |
| Cloudflare (R2, CDN, Turnstile) | File storage, content delivery, bot protection | Uploaded and generated media, IP address, request metadata | United States / global edge |
| Fly.io | Hosting for the varg API and render service | All API request and response content, including prompts and media | United States |
| Vercel | Web application hosting and product analytics | Request metadata, usage and performance data | United States |
| Upstash | Redis: rate limiting and generation result cache | Hashed request parameters, result references, usage counters | United States |
| Stripe | Payments, subscriptions and tax calculation | Email, billing address, payment and transaction data | United States |
| Resend | Transactional and waitlist email | Email address, name | United States |
| Sentry | Error monitoring and diagnostics | Error reports, stack traces, request context | United States |
| Anthropic | Chat assistant and planning | Chat messages and prompts | United States |
| Groq | Transcription, and automatic captioning and safety classification of uploads | Uploaded images and video frames, prompts | United States |
2. AI model providers
When you generate media, your prompt and any reference files you supply are transmitted to the provider that serves the model you selected. We send the material required to fulfil the request; we do not send your account, billing or payment details.
| Provider | Purpose | Data | Location |
|---|---|---|---|
| fal.ai | Image, video and audio generation; routes to many model families | Prompts, reference media, generation parameters | United States |
| Image and video generation (Veo, Nano Banana) | Prompts, reference media | United States | |
| Black Forest Labs | Image generation (FLUX) | Prompts, reference media | Germany |
| ElevenLabs | Speech and music generation | Text, voice parameters, reference audio | United States / United Kingdom |
| HeyGen | Avatar and talking-head video | Prompts, reference images, audio, likeness data | United States |
| Higgsfield | Image and video generation | Prompts, reference media | United States |
| Together AI | Model inference | Prompts | United States |
| Rendi | FFmpeg media processing | Media files | Israel |
| PiAPI | Aggregator routing to ByteDance Seedance and Seedream models | Prompts, reference media | Hong Kong SAR |
3. Models developed in China
Several models offered through the Services are developed by companies based in China, and one routing provider is based in Hong Kong. If you select one of these models, your prompt and any reference media — which may include photographs of identifiable people — are processed by or on behalf of that developer. This includes:
- Kling — Kuaishou Technology (China)
- Seedance, Seedream, SeedVR, OmniHuman — ByteDance (China), reached through fal.ai or PiAPI (Hong Kong)
- MiniMax / Hailuo — MiniMax (China)
- Wan, Qwen — Alibaba (China)
Data protection and government access laws in these jurisdictions differ materially from those in the EEA, the United Kingdom and the United States, and we cannot guarantee an equivalent standard of protection. If this is unacceptable for your use case, choose a model served from another jurisdiction and do not upload personal data as reference material.
4. Changes
We update this page when we add or remove a subprocessor. To be notified of changes, or to request the transfer mechanisms we rely on, write to hello@varg.ai.