100+ models
Frontier and open-weight catalogue, hosted in Australia.
Sovereign AI
Sovereign multimodel inferencing.
Onshore compute, security and control. 100+ frontier and open-weight models behind an Sovereign Australian inference.
100% Australian & Compliant by Design
ISO 27001Zero retentionNo training on your data
What it is Run any model. In Australia. Without changing your code.
Run any model. In Australia. Without changing your code.
In Australia.
Built for AI and ML workloads that need sovereign GPU compute. Run frontier and open-weight models in Australia under Australian contract.
Browse the model catalogue → Model catalogue
100+ models, hosted in Australia.
Open-weight and frontier models from Meta, Mistral, Qwen, DeepSeek, Anthropic and OpenAI, served from Sydney. One sovereign contract for all of them. AUD billing, zero retention, no cross-tenant caching.
Built for Australian AI workloads
Sovereign by design. Compatible by default.
Drop Amaze into your existing AI stack. Sovereign Australian inference, AUD pricing, and a sovereign Australian endpoint behind every model.
Zero retention
Prompts and outputs discarded by default. No tenant caching.
Burst GPU access
Scale into dedicated GPU pools when batch jobs demand it.
Carrier-grade transit
Up to 40 Gbps onto the Amaze backbone, predictable latency.
Enterprise grade network performance
99.9 %
Uptime SLA
Financially backed, measured monthly.
100 +
Models in catalogue
Frontier and open-weight, hosted in AU.
40 Gbps
Transit per region
Carrier-grade Australian backbone.
0
Cross-tenant caching
Zero prompt retention by default.
Catalogue
Run any open-weight or frontier model, in Australia.
All major LLM families, dozens of sizes, one Australian sovereign compute engine. Pick the model, call the endpoint, ship the feature.
Anthropic
OpenAI
Meta · Llama
Mistral AI
Google · Gemma
Qwen
DeepSeek
Hugging Face
Sovereign inference, built the way regulated Australian teams need it.
Five things make Amaze Multimodel Inferencing feel different from calling a hyperscaler endpoint that happens to live in Sydney.
Zero retention by default
Prompts and outputs are discarded once the response is served. Nothing cached across tenants, ever.
No training on your data
Customer data is never used as training data. Not by Amaze, not by the model vendor. Written into the contract.
Multi-node scale
Scale inference from a single endpoint to multi-node tensor-parallel deployments without rearchitecting. Same OpenAI-compatible API surface throughout.
Hybrid by design
Integrate sovereign inference with existing on-prem or hyperscaler workloads via IP Transit and VPC. The control plane stays Australian regardless of where the calling app sits.
Onshore data for open-source models, end-to-end
Inputs, outputs, logs and metadata stay in Australian regions for every open-weight model on the catalogue. Subject to Australian law, full stop.
Ready to run AI in Australia?
Spin up sovereign inference in minutes.
Amaze Multimodel Inferencing is generally available in Sydney (ap-syd-2). Reserved endpoints, burst capacity and dedicated fine-tunes all live on the same control plane.
Built on enterprise-grade infrastructure
Multimodel Inferencing, frequently asked questions
What is sovereign AI inference?
Sovereign AI inference means running trained models on Australian-controlled infrastructure, in Australian regions, under Australian law. With Amaze, prompts and responses never leave the country, your data isn't used for model training, and there's no prompt caching across tenants, the model serves your request and that's it.
Which models can I run through Amaze?
More than 100 frontier and open-weight models across Meta, Mistral, Google, DeepSeek, Qwen, Anthropic, OpenAI, Cohere, Stability and others. The catalogue spans hosted multimodel inferencing, dedicated GPU endpoints for fine-tunes, and tokens-as-a-service for bursty workloads.
How is Amaze different from a hyperscaler region in Sydney?
A hyperscaler region in Sydney is still operated under foreign law, with billing, support and the control plane offshore. Amaze is an Australian company running Australian infrastructure with Australian engineers on the support line, lower TCO, no egress between our regions, and contracts written under Australian jurisdiction.
Is Amaze suitable for regulated workloads?
Yes. The platform is ISO 27001 certified, hosted in Tier 3+ Australian data centres with carrier-neutral connectivity. Enterprise, education, software, AI/ML, finance and healthcare customers run sovereign workloads on Amaze today.
How do I move data into Amaze from another cloud?
The Amaze migration service covers the egress cost and provides a fully managed transfer, end-to-end validation, and cutover support. No account closures required at your existing provider, and no exit penalties from us if you ever choose to leave.
Run model inferencing securely onshore, with Australia's sovereign AI engine.
Talk to an Australian solution architect about onboarding. Burst capacity is available; reserved endpoints are too.