Powerful models.Unbeatable prices.European infrastructure.Zero data retention.
Serenity Edge serves open-weight models and the Orion family from our own infrastructure in Spain, through an OpenAI-compatible API, with per-million-token pricing and zero data retention.
Change the base URL. The rest of your code stays the same.
from openai import OpenAI
client = OpenAI(
base_url="https://api.serenityedge.ai/v1",
api_key="YOUR_API_KEY",
)
prompt = "Summarise this email in three bullet points: …"
reply = client.chat.completions.create(
model="orion-plus",
messages=[{"role": "user", "content": prompt}],
)
print(reply.choices[0].message.content)A real answer from orion-plus, recorded on 14 October 2026.See the full request
- Issue: Order 8812 arrived with two damaged boxes.
- Request: Urgent replacement of the damaged items is required before Friday to meet a Saturday trade fair deadline.
- Alternative: If replacements cannot be delivered by Friday, a refund for the two boxes is requested.
Why Serenity Edge
Three reasons to choose Serenity Edge
EU data, zero retention
Inference runs on our own infrastructure in Spain and processing never leaves the European Union. Prompts and completions are never written to disk or logs.
Per-token pricing, in plain sight
A price per million tokens, published on this site for every model in the catalogue. No minimums, no lock-in and no surprises on the invoice.
OpenAI compatible: change one line
The SDKs and tools you already use work as they are: add your key and keep working with streaming, tools and structured outputs.
Catalogue
The whole catalogue through one API
Orion, Serenity’s own family, and open models served with their original weights, to generate text, read documents, make decisions, search and redact. Every model with its price published per million tokens.
Chat and vision
Text generation from text and images, with streaming, tools and structured outputs.
- Orion Pro
- Orion Plus
- Qwen3.8 27B
- Qwen3.6 35B A3B
POST /v1/chat/completionsInput from 0.101 USD per million tokens
Document OCR
Text, tables, formulas and chart data extracted from the image of a page.
- PaddleOCR-VL 1.6
POST /v1/chat/completionsInput from 0.016 USD per million tokens
- Preview
Decisions
Typed answers with their probability, to route, classify, filter and triage.
- Orion Compass
- Orion Compass Flash
- d3
Embeddings
Text turned into vectors for semantic search, clustering and RAG.
- Qwen3 Embedding 0.6B
Rerank
Documents ranked by their relevance to a query before they reach the model.
- Qwen3 Reranker 0.6B
Redaction
Names, email addresses, phone numbers and other personal data found and masked in text.
- Privacy Filter
Prices and models read from the API on 6 Nov 2026, 19:46 UTC, when this page was published.
Our decision model
Play against Orion Compass
On every turn it picks its move with a real call to the Serenity Edge API. Change its board and watch it decide again.
- A real API call
- The probability of each move
- Nothing scripted
And four business cases on the same API
Getting started
In production in three steps
curl https://api.serenityedge.ai/v1/chat/completions \ -H "Authorization: Bearer $SERENITY_EDGE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "orion-plus", "messages": [{ "role": "user", "content": "Summarise this email in three bullet points: …" }]
}'A real answer from orion-plus, recorded on 14 October 2026.
"usage": {
"prompt_tokens": 81,
"completion_tokens": 66,
"total_tokens": 147
}Everything the API includes
Streaming responses
Tool calling- Structured outputs (JSON)
- Image input and document OCR
- Decisions (preview)
- Embeddings
- Rerank
- Personal data redaction
The decision, embeddings, rerank and redaction routes need a Serenity Edge API key.How to get one
Infrastructure
Our own inference nodes in Spain
Serenity Edge does not rely on third-party clouds for inference: models run on our own inference nodes, located in Spain and operated by a Spanish company. A single API gives access to the whole catalogue and your data is processed within the European Union.
- Your application
- api.serenityedge.ai
- Our own nodes in Spain
- Response
Security and compliance
Designed so that your data never leaves the EU
Zero data retention
Prompts and completions are never written to disk or logs, and are never used for training.
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.
Status and availability
Transparent about availability too
Availability and incidents on a public status page. Uptime is measured from outside our infrastructure, not declared by us.
Each catalogue category has its own monitor, testing it with real requests every few minutes: chat and vision, document OCR, decisions, and embeddings, rerank and redaction. If something fails, you can see where in the catalogue.
See service statusWays to access
Choose how to start
Serenity Star account
Self-service: create the account, generate your key and buy credits. The direct route for teams and developers.
When you create a Serenity Star account you get a welcome credit to try the chat models.
The decision, embeddings, rerank and redaction routes need a Serenity Edge API key.How to get one
Third-party platforms
Use Serenity Edge as a provider from your cortecs or EUrouter account, billed by that platform. More platforms are on the way.
See the platformsContract with SLA
Dedicated capacity, specific terms and a service level agreement for organisations.
Contact usFrequently asked questions
What people usually ask us
Do you retain my prompts or completions?
No. Prompts and completions are never written to disk or logs. So that repeated prefixes can be reused across requests, prompt prefixes may stay in a memory cache for less than 24 hours, after which they are discarded automatically; they are never used for training.
Where is my data processed?
Inference runs on our own infrastructure in Spain and all processing takes place within the European Union. We do not transfer your data outside the EU.
How does billing work?
Per million tokens, with no minimum fee and no lock-in. Chat models bill input and output separately, with cached input at 25% of the input price; decision, embeddings, rerank and redaction models bill input only. With a Serenity Star account you buy credits and use them; through cortecs or EUrouter, that platform bills you; contracts have their own terms.
What are the limits and what happens if I exceed them?
Default limits are generous and are raised under contract. If you exceed your limit, or the API cannot take the request at that moment, it responds with status 429; retry with an increasing delay, as with any OpenAI-compatible API.
What is the difference between Orion and the open models?
Open models are served with their original weights and their usual identifier. Orion is Serenity’s own model family: Orion Pro for demanding tasks, Orion Plus for large volumes and, in preview, the Orion Compass decision models. Both families share the API; the context, capabilities and price of each model are published on the models page.
Which platforms can I use Serenity Edge from?
From cortecs and EUrouter, as a provider, with each platform’s account and billing, or directly with a Serenity Star account or a contract for organisations. More platforms are on the way, OpenRouter among them.
Do you offer an SLA?
Only under contract. The public terms of service are based on commercially reasonable efforts, and service status is published on an open page.
Start today with your API key
Open-weight models and Orion, published per-token prices and EU data, on the API you already use.