AI models, served in the EU.

• Internal assistants over company documents.

  Connect your own retrieval system and ask questions over HR policies, legal templates and operating procedures. Your systems keep the source documents; Zarx handles the model request.

• Customer support with personal data.

  Summarize tickets, classify intent and draft replies for workflows that contain names, accounts and contracts. Keep the customer records in your systems and send only what the request needs.

• Structured extraction.

  Turn invoice and contract text into structured JSON your systems can validate. Use the Responses API and strict schemas for repeatable outputs.

• Agents and tool calling.

  Let the model reason on Zarx and call your backend through function tools. Your tools and databases stay on your side.

• Batch and evaluations.

  Upload a file of requests and collect the results later. Use the same models for backfills and evaluations without keeping someone waiting.

• Replacing a provider outside the EU.

  Point your existing integration at Zarx and choose a model from the catalog. Your code keeps its request shapes while inference runs in the EU.

Tell us your workload

curl -N https://api.zarx.ai/v1/chat/completions \
  -H "Authorization: Bearer $ZARX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "z-pro",
    "messages": [
      {"role": "user", "content": "Where is this request processed, and under which law?"}
    ],
    "stream": true
  }'

from openai import OpenAI

client = OpenAI(base_url="https://api.zarx.ai/v1", api_key=ZARX_API_KEY)

with client.responses.stream(
    model="z-ultra",
    input="Extract invoice number, supplier, date and total from this invoice: ...",
    text={"format": {
        "type": "json_schema",
        "name": "invoice",
        "schema": {"type": "object", "properties": {
            "invoice_number": {"type": "string"},
            "supplier": {"type": "string"},
            "date": {"type": "string"},
            "total": {"type": "number"}
        }, "required": ["invoice_number", "supplier", "date", "total"], "additionalProperties": false}
    }},
) as stream:
    for event in stream:
        if event.type == "response.output_text.delta":
            print(event.delta, end="")

Sample output. The requests are in the shapes the API accepts; the replies are illustrative. The base URL is the design target for early access; access is by invitation.

Operated by NPAW.

A Barcelona company that has run its own data infrastructure for streaming services since 2008, and runs this inference gateway in production for its own products.Who we are

Your jurisdiction, by design.

Inference on NPAW’s own GPUs in the Barcelona area (Spain), API entry on NPAW hardware in Haarlem (Netherlands), both operated by NPAW under EU law, with no US sub-processor on the data path. Designed to help you keep inference under EU jurisdiction and assess GDPR data-residency requirements.The data path

We publish what we can prove.

No latency, throughput or price figure appears on this site without evidence you can check. Every statement traces to a source, including the certifications NPAW holds: ISO 27001, ISO 27701, ISO 42001 and ENS High.What we do not do yet

Sources: NPAW infrastructure inventory, 2026-09-03 (server-by-server record of NPAW-operated GPU hosts); NPAW infrastructure inventory (GPU hosts in the Barcelona area; gateway and API entry on NPAW hardware in Haarlem, Netherlands); legal entity NicePeopleAtWork S.L. (npaw.com legal notice); GDPR territorial scope, Art. 3; NPAW compliance page, verified October 2026.