Egen API. Open models in Sweden, one API.

Chat, embeddings, reranking and Swedish speech to text, through an OpenAI-compatible API on servers in Sweden. What you send is not stored, and nothing is trained on it. We are finishing the setup: join the waiting list and we will write when your access is ready.

Join the waiting list Read the quickstart

curl https://api.quince-ai.com/v1/chat/completions \
  -H "Authorization: Bearer $EGEN_API_KEY" \
  -d '{"model": "gemma-4-31B-it",
       "messages": [{"role": "user",
         "content": "Summarise the attached minutes in three bullet points."}]}'

1. The board approved the Q3 budget. 2. …

What it does

Five kinds of call. One address.

Egen API speaks the same language as the OpenAI API, so the libraries and tools you already use connect by changing two settings: the address and the key.

Endpoints

POST /v1/chat/completions: chat, with streaming, tool calls and JSON output
POST /v1/embeddings: vectors for search and retrieval
POST /v1/rerank: sort search results by relevance
POST /v1/audio/transcriptions: speech to text, tuned for Swedish and Norwegian
GET /v1/models, POST /v1/tokenize and GET /v1/usage: the model list, token counts and your usage

For your own tools

Works with the OpenAI libraries for Python and JavaScript, and with any tool that accepts an OpenAI-compatible address and a key
Keys belong to a project in your organisation, are shown once and can be revoked at any time
Usage per request and per project in the console
An Idempotency-Key header stops a retry from starting a second generation

Plans

Four plans. Pay for what you use.

Each plan is a monthly top-up of credit, spent at a list price per model. Usage above your credit is charged on top, and unused credit rolls over to the next month. Prices on request while we finish the setup.

Trial

For trying it out

Starting credit to try it

Basic rate limits
Support by email
No monthly cost

Starter

For a small team

Credit each month

Higher rate limits
Support by email
Automatic billing by card

Team

For development teams

Credit each month, 10% extra

Higher rate limits
Priority support
Automatic billing by card

Enterprise

For production at scale

Credit each month, 10% extra

Highest rate limits
Dedicated support
Invoice and Peppol billing

Rate limits are shared per account. Cached input tokens are billed as regular input tokens. All models →

What happens to your data

Prompts and answers are not stored, and nothing is trained on them
Usage records (time, model, token counts) are kept for 90 days
The models run on servers in Sweden that we operate
No US company handles what you send

Billing

A list price per model, per million tokens or per 1 000 minutes of audio
Card payment on Trial, Starter and Team; invoice and Peppol on Enterprise
Failed requests and missing usage are never charged
The same data processing agreement as the Egen workspace

Is Egen API the same as the Egen workspace?

No. The workspace is a finished tool for your whole team at a flat monthly price. Egen API gives your own systems the models directly, and you pay for what you use.

Who made the models?

Every model is open weight, published on Hugging Face by its maker: Google, Mistral AI, Qwen, Z.ai, Moonshot AI, BAAI, KBLab, NbAiLab, Systran, Klang and intfloat. We run them on our own servers in Sweden, so none of those companies sees what you send.

Is what I send stored?

No. Prompts and answers are not stored and not trained on. We keep usage records, such as time, model and token counts, for 90 days for billing and support.

When can we start?

We are finishing the setup. Join the waiting list, tell us what you want to build, and we will write when your access is ready and send the prices.

What happens to unused credit?

It rolls over to the next month.

Egen API

Join the waiting list.

We are finishing the setup. Tell us what you want to build, and we will write when your access is ready, with the prices.