Subprocessors.
Synapse Garden uses a small set of carefully chosen third-party services to deliver the product. Each one handles a specific job and is bound by a data processing agreement. We update this list when we add or remove a subprocessor.
- VercelHosting, edge runtime, and AI Gateway routing for upstream model providers.Global
- SupabaseAuthentication, Postgres database, and managed user identity. Stores account email, hashed credentials, and dashboard data.us-east-1 (US)
- UpstashRedis used for API key validation cache and per-project rate limiting.Global (low-latency edge)
- Dodo PaymentsSubscription billing, plan management, invoices, and one-time overage charges.Global
- ResendTransactional email delivery — verification, magic links, password reset, billing receipts.Global
- SentryError monitoring and unhandled exception telemetry. PII redaction is enabled.Global
When you make a request to Synapse Garden, it is forwarded — through Vercel AI Gateway as our routing partner — to the upstream model provider you selected. Customer data sent in a request reaches that provider only. Providers we currently support include:
- · OpenAI
- · Anthropic
- · Meta
- · Mistral
- · DeepSeek
- · xAI
- · Cohere
- · Alibaba (Qwen)
- · Perplexity
- · Moonshot AI
- · MiniMax
- · NVIDIA
- · ByteDance
- · And other model providers available through the routing layer
Each provider has its own privacy and data-retention policy. Your request body is never used to train provider models when the underlying API exposes a no-train flag (we set it where available). By default, Synapse Garden does not retain prompt or completion content — only token counts, model id, latency, and status. Full payload logging is opt-in at the org level.
We update this page within 30 days of adding or replacing a subprocessor. If you operate under a contract that requires earlier notice, email legal@synapse.garden and we will subscribe you to direct change notices.
Related: Pricing disclosure · Privacy policy · Terms of service