New

Sandboxes, in private preview: isolated microVMs for AI agents, compatible with the E2B SDKs.Read more

Serenity Edge

Security and data residency

Your data, under your control and within the EU

Serenity Edge is designed so that you can use language models without giving up control of your data: zero retention, processing within the European Union and published legal documents.

The path of your data

From your application to the model and back, inside the EU

What happens to your request at each step, and what never does. Each guarantee links to the clause of the legal documents that states it.

  1. Your application
  2. European Union

    1. api.serenityedge.ai
    2. ModelOur own inference nodes
  3. Answer
From your application to the model and back, inside the EUYour application sends a request with your API key. It enters the European Union, goes through the Serenity Edge API and reaches the model, on our own inference nodes; the answer goes back to your application. On the way, nothing is written to disk or logs, the prompt prefix stays in a cache for your key only and nothing is used for training.European UnionAnswerYour applicationapi.serenityedge.aiModelOur own inference nodes1234
  1. Processed in the EU

    Request handling and inference take place in the European Union, the models on our own infrastructure.

    Data processing agreement, clause 7.1
  2. Nothing on disk or in logs

    Prompts and completions are processed in memory, never written to disk or logs, and not inspected by staff.

    Terms of service, section 6
  3. A cache for your key only

    Repeated prompt prefixes may stay in a volatile memory cache for at most 24 hours. That cache is never shared across API keys.

    Data processing agreement, clause 9.2
  4. Never used for training

    Your prompts and completions are not used to train, fine-tune or evaluate any model, Orion or open.

    Data processing agreement, clause 2.2
Read data residency and retention in the documentation

Zero data retention

Prompts and completions are never written to disk or logs. So that prefixes repeated with the same API key can be reused, prompt prefixes may stay in a memory cache for at most 24 hours, after which they are discarded automatically; they are never shared across API keys and never used for training. Zero retention applies to inference: a sandbox keeps your files while it exists, and a paused one keeps its memory and disk for up to 30 days.

Inference in Spain, processing in the EU

Models run on our own infrastructure in Spain and your prompts and completions are processed within the European Union. Requests sent from the EU stay in the EU.

Encryption in transit

All traffic to the API is encrypted with TLS.

Minimal metadata

We keep only the metadata needed for billing and abuse prevention, such as token counts and timestamps, for 90 days. It does not include request content.

GDPR and published documents

Substrate Artificial Inteligence, S.A. is a Spanish company subject to the GDPR. The terms of service, privacy policy, data processing agreement and list of sub-processors are published and version-controlled.

Commitments under contract

Service level agreements (SLAs) are established only under contract. The public terms are based on commercially reasonable efforts and service status is public.

Availability in plain sight

Service status and incidents are published at status.serenityedge.ai, with uptime measured from outside our infrastructure.

See service status

Need specific terms?

For a tailored data processing agreement, dedicated capacity or an SLA, let’s talk.