One balance, every model
Top up once in EUR or USD. Spend the same balance on any model in the catalogue — no per-provider minimums, no idle prepayments stranded with a vendor you stopped using.
EURU is a European inference gateway. Buy tokens once and spend them across models from Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek and more — a single endpoint, a single balance, a single invoice from an EU-registered company.
# Any model, same request shape curl https://api.euru.io/v1/chat/completions \ -H "Authorization: Bearer $EURU_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "anthropic/claude-sonnet-5", "messages": [ { "role": "user", "content": "Summarise this contract." } ] }'
# Works with the official OpenAI SDK — just change the base URL from openai import OpenAI client = OpenAI( base_url="https://api.euru.io/v1", api_key=os.environ["EURU_API_KEY"], ) resp = client.chat.completions.create( model="openai/gpt-5.6-sol", messages=[{"role": "user", "content": "Hello!"}], ) print(resp.choices[0].message.content)
import OpenAI from "openai"; const euru = new OpenAI({ baseURL: "https://api.euru.io/v1", apiKey: process.env.EURU_API_KEY, }); const res = await euru.chat.completions.create({ model: "google/gemini-3.8-flash", messages: [{ role: "user", content: "Hello!" }], // automatic failover if the primary is unavailable fallbacks: ["anthropic/claude-sonnet-5"], });
Routing to
Every provider has its own contract, its own credit card charge, its own SDK and its own rate limits. EURU collapses all of it into one commercial and technical interface.
Top up once in EUR or USD. Spend the same balance on any model in the catalogue — no per-provider minimums, no idle prepayments stranded with a vendor you stopped using.
The API speaks the OpenAI chat-completions dialect. Point your existing SDK at api.euru.io/v1, change the model string, and you are done — streaming and tool calls included.
Declare a fallback chain per request. If an upstream is rate-limited, degraded or down, we re-route to the next model in your list instead of returning a 429 to your users.
Per-request cost, token counts and latency on every response header, and in the dashboard. Attribute spend to customers, features or environments with API-key level tagging.
Contract with an EU-registered entity, invoiced in EUR with a valid VAT number. GDPR-aligned data terms, a signable DPA, and EU-region routing where the upstream supports it.
Hard limits per key, per model and per day. Nothing runs away overnight, and a leaked key cannot cost more than the ceiling you set for it.
Pay by card, SEPA transfer or invoice. Your balance is denominated in credits at a fixed 1:1 rate to USD list price, with EUR settlement available.
Issue as many keys as you need — one per environment, per customer or per feature — each with its own spend ceiling and model allow-list.
Swap your base URL, keep your SDK. Prefix the model name with its provider and start sending traffic. Usage is metered per token as it streams.
Our price in USD per million tokens — 5% below the market reference rate.
| Model | Provider | Context | Input / 1M | Output / 1M |
|---|
Start with a €50 top-up instead of a vendor onboarding cycle. Evaluate five models in an afternoon and keep the one that wins on cost per resolved ticket.
Route cheap traffic to open-weight models and reserve frontier models for the hard 5%. Per-key attribution shows exactly which feature is burning the budget.
A single EU supplier, one purchase order, one monthly invoice. Add a new model to your stack without another procurement, security review or DPA.
EURU is operated by Paytend Europe UAB, a company registered in Lithuania. You contract with an EU legal entity, receive VAT-compliant invoices, and sign a data processing agreement governed by EU law.
We are a payments-and-platform company by background, so metering, reconciliation and audit trails are treated as first-class product surface rather than an afterthought.
Documented sub-processors, EU standard contractual clauses where transfers occur, and a signable DPA.
Prompts and completions are never used to train models, ours or anyone else's. Logging is opt-in and retention is configurable.
Every request is billed against upstream-reported token counts, exportable as CSV for reconciliation against your own logs.
Get a key, top up whatever you are comfortable with, and route your first request today. Talk to us when you need volume pricing or a signed agreement.