AI Inference On NZ GPUs Billed in NZD

OPENAI-COMPATIBLE LLM API • WORKS WITH ANTHROPIC CLIENTS NO TRAINING ON YOUR PROMPTS OR DATA

Change one base URL and your existing SDKs and agent tooling call open-weight models running on ASI GPUs in New Zealand. Turn it on, create a key and see spend in real time, with prepaid NZD credits and no exchange-rate drift. Sign up with your name, email and a password. You add a card before you start using a service.

1 URL

To change in your existing client

NZD

Prepaid credits, no exchange-rate drift

Zero

Training on your prompts or data

Per-key

Budgets, rate limits and spend caps

Compatibility

Drop-in API

OpenAI-compatible, and compatible with Anthropic clients. Change one base URL and existing SDKs and agent tooling work unmodified.

Hosting

Runs on NZ GPUs

Open-weight models running on ASI GPUs in New Zealand. Your prompts and data are not used to train models.

Billing

Prepaid NZD credits

Billed in NZD with no exchange-rate drift. Pay per token, with input, cached input and output metered separately.

Controls

Per-key limits

Create keys for each app or team, each with its own budget, rate limit and spend cap. See spend in real time.

Models

Curated models

Curated instruct, reasoning, vision and embedding models, all running on ASI GPUs in New Zealand.

Tiers

Basic and Enterprise

The Basic tier is self-service, for prototyping and internal tools. For production with reserved capacity, the Enterprise tier is contracted.

How it works, in minutes

No quote, no sales call, no ticket.

Everything happens in the portal at cloud.asi.nz, in one session.

It works the same for solo developers, freelancers and teams of any size: sign up with your name and email, add a card, buy prepaid NZD credits and create a key.

1. Sign up Sign up with your name and email at cloud.asi.nz.
2. Add a card Add a payment method and buy prepaid NZD credits.
3. Create a key Turn on AI Inference, create a key and copy the secret, which is shown once. Point your OpenAI or Anthropic client at the base URL shown in the portal.

ai-inference: zsh

# Key and base URLs from the portal
export OPENAI_API_KEY="$ASI_INFERENCE_KEY"
export OPENAI_BASE_URL="openai-base-url-from-portal"

# Any OpenAI-compatible client works unchanged
curl -s "$OPENAI_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "model-slug",
       "messages": [{"role": "user", "content": "Kia ora"}]}'

# Anthropic clients use their own base URL
export ANTHROPIC_BASE_URL="anthropic-base-url-from-portal"

Account

Can I sign up as an individual?

Yes. Register with your name and email, and that becomes your account. Add a card and buy prepaid NZD credits, then create a key. You get the same NZ-hosted service and NZD billing as any organisation.

Data

Where do my prompts go?

To open-weight models running on ASI GPUs in New Zealand. Your prompts and data are not used for training. Account and billing data and card payments can be processed offshore.

Cost

How is it billed?

In NZD, from prepaid credits, so there is no exchange-rate drift. Input, cached input and output tokens are metered separately, and per-key spend caps stop runaway usage.

Code

Do I need to rewrite my code?

No. It is OpenAI-compatible and works with Anthropic clients. Change the base URL and key, set the model name, and existing SDKs and agent tooling work unmodified.

Production

Can I run production on it?

The Basic tier suits prototyping and internal tools. For production with reserved capacity, the Enterprise tier is contracted: book a meeting to set it up.

Call an LLM API Running in New Zealand

OpenAI-compatible, works with Anthropic clients, billed in NZD. Turn it on, create a key and see spend in real time.

Want help sizing it, or reserved capacity for production? Book a meeting

All four products, one account