NewEco lane for GPT-6 Astra: the same model at a lower price

Currency

You always top up in euro. Other currencies are shown at today's ECB rate. ECB reference rates

Sign inGet started
NexroutingEvery request, the right model.

One API for every AI model.

Text, code, image and video models from OpenAI, Anthropic, Google, xAI and more, plus GPUs by the hour. One key, one balance, paid in euro.

  • 148 models, one keyWorks with the OpenAI and Anthropic SDKs. Change two lines.
  • Pay in euroiDEAL, Bancontact or card. No subscription, and a VAT invoice with every top-up.
  • Up to 40% below the official priceThe same model on the Eco lane. Standard costs what the provider charges.
  • GPUs from €0.26/hourH100, H200, B200 and more. Billed per minute.
Company in the EUGDPR and DPAPublic status pageiDEALBancontactCard
Models from
AnthropicOpenAIGooglexAIDeepSeekQwenMoonshot AIZ.aiMiniMaxByteDanceTencentXiaomiBlack Forest LabsKling
Savings

Pay less than the provider's own price.

Standard costs exactly what the provider charges. The Eco lane runs the same model for up to 40% less.

−40%GPT-6 AstraLive€9.12 → €5.47 · input per 1M−29%Gemini 3.1 Pro Preview€1.82 → €1.29 · input per 1M−40%GPT-6 Sol€1.82 → €1.09 · input per 1M−29%Gemini 3.8 Flash€0.68 → €0.48 · input per 1M−40%GPT-5.5€4.56 → €2.74 · input per 1M
Savings

What you save each month

EUR
The model you use most
Share of traffic on Eco
Eco is the same model with lower priority when it is busy. Batch work and background jobs usually fit there.
You save per month€200€2,400 a year · 40% less than the official price

We compare with the provider's own official price. Standard costs the same; the saving comes from the Eco lane.

Get started
Two products, one account

Models and GPUs in one account.

One balance in euro pays for both. One invoice, one dashboard, one team.

NexPodsComing soon

A GPU by the hour, from the same balance.

Start a pod with vLLM, ComfyUI or Jupyter ready. Stop it and only storage is billed.

7GPU types
1 minbilling per minute
−45%below RunPod, per hour
H100 80 GB€2.73/h
H200 141 GB€3.36/h
B200 192 GB€5.28/h
RTX 4090 24 GB€0.45/h
See GPUs
What you get

Everything around the API, built in.

Keys, credit, usage and invoices in one place, for every model in the catalog.

One key

Every model behind one key.

Switch models by changing one string. Your key, your credit and your logs stay the same.

client.chat.completions.create(    model="anthropic/claude-opus-5.5",    messages=[{"role": "user", "content": "Hello"}],)
200 OKanthropic/claude-opus-5.51.0 s€0.0033
148models
14providers
1key and one balance
Lanes

Standard or Eco.

Eco is the same model at a lower price, with lower priority when it is busy.

gpt-6-astra · input per 1M
Standard€9.12
Eco€5.47
Pay in euro

Top up with iDEAL.

iDEALBancontactCard
Top-up€25.00CreditUS$28.78Invoicewith VAT
Open

See every request.

Which model answered, how many tokens, how long it took and what it cost.
15:41:58anthropic/claude-opus-5.51,240 → 380€0.0115
15:41:51openai/gpt-6-astra:eco3,200 → 640€0.0350
15:41:40google/gemini-3.8-flash820 → 210€0.0013
15:38:07openai/gpt-6-astra940 → 0free
Budgets

A budget per key.

Give each key a monthly budget, a list of models and an expiry date.

sk-nx-… production€283 / €456
sk-nx-… staging€8 / €46
sk-nx-… support agent€100 / €109
Fallback

If a route fails, we take the next one.

Same model, same price. Failed requests are not charged.

route 15020.4 s
route 22001.2 s
openai/gpt-6-astra · 1 answerbilled once: €0.0584
Keep your code

Change two lines. Keep the rest.

Point the OpenAI or Anthropic SDK at Nexrouting and use your Nexrouting key. Everything else in your code stays the same.

OpenAI SDKAnthropic SDKLangChainn8nCursorcurl
Read the quickstart
import osfrom openai import OpenAI client = OpenAI(    base_url="https://api.nexrouting.com/v1",    api_key=os.environ["NEXROUTING_API_KEY"],) reply = client.chat.completions.create(    model="anthropic/claude-opus-5.5",    messages=[{"role": "user", "content": "Hello"}],)
From sign-up to first request

Three steps to your first request.

This is the dashboard. Add credit, create a key, paste the code.

Pricing

Pay for what you use.

Open the calculator
google/gemini-3.1-flash-lite€0.75Classify 10,000 support tickets.10,000 × (300 in + 5 out tokens)
anthropic/claude-haiku-5.5€0.08Summarize 1,000 tickets in two lines.1,000 × (412 in + 96 out tokens)
anthropic/claude-opus-5.5€10.94Write 100 long reports from source documents.100 × (20,000 in + 2,000 out tokens)
Built in Europe

Built in Europe. Billed in euro.

Pay in euro

Top up with iDEAL, Bancontact or card. Credit in dollars at the ECB rate, so you see the exact amount first.

VAT invoices

Every top-up comes with an invoice. EU businesses with a VAT number get the reverse charge.

GDPR and a DPA

A data processing agreement ready to sign, and a public list of who processes data for us.

Keys stored hashed

You see a key once. We keep a hash, never the key itself.

Two-factor sign-in

Passkeys and 2FA for every account, and roles for your team.

Public status page

Uptime per model and per service, with every incident written up.

Questions

What people ask first.

How does billing work?

You top up in euro with iDEAL, Bancontact or card. We add credit in US dollars at today's ECB rate minus a 1% buffer. Every request is taken from that credit, and every top-up comes with an invoice that shows VAT.

Do I need to change my code?

Two lines: the base URL and the key. The API follows the OpenAI and Anthropic formats, so the SDKs you already use keep working.

What is the difference between Standard and Eco?

Eco is the same model at a lower price. When a model is busy, Standard requests go first. Add :eco to the model ID to use it, where a model offers it.

What happens when a route fails?

We retry on another route for the same model and the same price. Requests that fail are not charged.

Can I get my money back?

Unused credit, yes. Ask us from your account and we refund it to the payment method you used.

Do you offer GPUs?

Soon. NexPods lets you rent a GPU by the hour from the same account and the same credit, billed per minute.