🇪🇺 Developer API — OpenAI-compatible

The Depaza API. Drop-in OpenAI compatible.

Point the official OpenAI SDK at Depaza and get three EU-hosted models with web search and tools built in. Prepaid, pay-as-you-go — no subscription.

Drop-in compatible

Use the OpenAI SDK you already have — just change the base URL and key. /v1/chat/completions and /v1/models, streaming or buffered.

Strong models + tools included

Core is the capable baseline. Max and BOSS for deeper work. All with server-side web search and live page reading — answers grounded in the live web, not just the model.

You can see what you spend

A key per purpose, usage and cost per key, and one-click revoke. Nothing runs up a bill you cannot account for.

EU First

Models run in EU regions, GDPR-ready, covered by our standard DPA. The supporting services we use — payments, transactional email, CDN and optional web search — are disclosed at depaza.eu/subprocessors.

Your account

A console built for a company, not a login

Keys, spend and access belong to the organization. Invite a colleague and they see the same account — not a copy of it.

A key per purpose

Name each key for what it does, see who created it, and revoke one without touching the rest.

Usage you can read

Spend, requests and tokens over any period, broken down by model — and by key.

One balance, one company

Colleagues draw on the same prepaid credit. Nobody has to expense their own API bill.

An audit trail

Every key created, renamed or revoked is recorded with who did it and when.

Create your API Console account

Up and running in two minutes

Create a key in Settings → API, then call the API with any OpenAI-compatible client.

curl https://depaza.eu/v1/chat/completions \
  -H "Authorization: Bearer dpz_live_…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "core",
    "messages": [{"role": "user", "content": "What changed in EU AI rules this week?"}]
  }'

Full API reference →

Transparent, usage-based pricing

You pay our underlying cost plus a flat markup — no hidden multipliers. Prices below are live.

Model Input / 1M tokens Output / 1M tokens
Depaza Lite lite €0.42 €1.31
Depaza Core core €0.56 €1.67
Depaza Max max €4.86 €9.72
Depaza Advanced advanced €4.86 €12.50
BOSS (Unrestricted) boss €2.78 €11.11
Web search €0.022 per external search

Model ids & how to call them (API docs) → BOSS mode →

Prices in EUR. Web search is only billed when a query reaches the external engine — answers served by our own index or cache are free. Token usage is billed at the rates above.

Benchmarked

Same coding task. ~70% lower cost.

We ran identical, oracle-graded tasks through Depaza and a leading US frontier API. Matching correctness — at roughly a third of the cost.

See the benchmarks →

Editors & IDEs

Use Depaza in Cursor — and any OpenAI-compatible IDE

Point Cursor, Cline, Continue or Zed at Depaza and code with EU-hosted models. Create a subscription key in Settings to draw from your plan’s weekly tokens, or a pay-as-you-go key to bill prepaid credit.

Base URL

https://depaza.eu/v1

Models

core · max · advanced · boss

Core (1×) · Max (2.4× tokens) · Super Advanced (3× tokens) — metered per model.

Model ids & how to call them (API docs) →

Chat, inline edit and agent (tool-calling) requests route through your key. Some editor-proprietary features (e.g. Cursor Tab) may not use a custom endpoint.

Create your API Console account

Where your data goes

EU-hosted, and specific about it

  • Models run in EU regions. Database, storage and backups in EU regions.
  • TLS 1.3 in transit, AES-256 at rest.
  • Our standard data processing agreement applies automatically to business customers. Countersigned copies within one business day on request.
  • Every sub-processor is published, with 30 days' notice before we add one.

We may use prompts to improve our own EU-hosted models, and that training stays in EU regions. Enterprise plans are excluded by contract.

Security & compliance → Data processing agreement → Sub-processors →

Start building today

Create an account, get a key in under a minute, and call the API with any OpenAI-compatible client. You only pay for what you use.