Skip to content
Documentation

The Cloud Virtualization Platform

All your infrastructure.
One cloud.

Bare metal in your rack, the cloud accounts you already pay for, and the hyperscalers’ regions — compute, networking, identity and services — answer as one cloud. You build against one thing instead of five.

The Hanzo console: machines, clusters and services from bare metal and connected cloud accounts, in one list.

Ten layers. One origin.

Every one of them answers on api.hanzo.ai/v1, to the same key, against the same balance. Reaching for the next layer is a line of code, not a procurement cycle.

The 10 layers of the Open AI Cloud, assembling base to crown: Web3, Compute, Data, Network, Security, Infrastructure, Observe, Dev, AI, Apps.
  1. 10

    Apps

    Production apps built on the cloud.

    Chat · Bot · Marketplace · Search · Crawl · Studio · Console

  2. 09

    AI

    Models, agents, and inference — open weights, one API.

    Models · Providers · Agents · Inference · Embeddings · Playground · Prompts

  3. 08

    Dev

    Clients, SDKs, and keys for the whole API.

    CLI · SDKs · API · Integrations · IDE · Desktop · API Keys

  4. 07

    Observe

    Logs, metrics, traces, and cost in one pane.

    Overview · Evals · Logs · Metrics · Traces · AI Metrics · Dashboards · Alerts · Billing · Plans & Pricing · Datasets · Experiments · Sessions · Scores · Score Configs · Annotation Queues

  5. 06

    Infrastructure

    Source to production as declared state.

    Projects · Environments · Builds · Registry · Releases · Pipelines

  6. 05

    Security

    Identity, keys, and audit for the whole cloud.

    IAM · Authz · KMS · Secrets · Audit · Zero Trust

  7. 04

    Network

    Connect, route, and protect every service.

    Gateway · Nodes · VPC · DNS · Load Balancer · Service Mesh

  8. 03

    Data

    Every persistence primitive — at rest and in motion.

    Vector · SQL · KV · Object Storage · Datastore · Base · DocDB · Memory

  9. 02

    Compute

    GPUs, machines, and functions that scale to zero.

    GPUs · Machines · Functions · Edge · Applications · Clusters · Kubernetes · Tasks

  10. 01

    Web3

    On-chain infrastructure for the cloud.

    Settlement · Wallets · Referrals · Tokens · Networks · Indexer · Oracles

Point at a layer.

Each one stands on the ones beneath it and answers on the same origin, to the same key. Pick the one you need and it tells you what it gives you and what is in it.

Layer

Apps

Production apps built on the cloud.

All apps

Every workload, one compute plane

A container, a whole machine, or a leased shell — the isolation is a choice per workload, not three products to buy.

The containers surface on api.hanzo.ai/v1/functions, operation by operation.

The assembly tax.

Nobody sends you a bill for it. You pay it in engineering — in the accounts, the keys, the SDK bumps, the security reviews, and above all in the code that holds one vendor's answer to the next vendor's input. It is the largest thing most teams buy, and it is invisible.

Per year, foreverAssembledHanzo Cloud
Accounts to open101
Keys to store and rotate101
SDKs to keep on a version101
Consoles to learn101
Invoices to reconcile101
Vendors to review for security101
Code holding it togetheryoursnone

Integrated means the joins are already made. One origin, one identity, one balance — you did not build them, and you do not maintain them.

One key. Twelve calls. No glue.

Models, routing, agents, sandboxes, memory, a channel, a site, a storefront, and the bill for all of it — one organization, one bearer token, one origin. Every operation below is one the platform serves: each is checked against the published OpenAPI document on every build, and a beat the document does not carry is dropped rather than drawn.

One org, one afternoon: a spaceship shop that did not exist this morning.

GET /v1/models

Returns the list of available models from the routing table.

Models

Reliable platform.

What we run for you is an image you can run yourself — the same artifact, on your own cluster or on a laptop. A platform you can hold is a platform that cannot strand you.

Live status, in public →

Support in the open.

The source is public, so an answer can be a commit and a bug can be fixed in the open — not filed with someone who has never read the code. And if it has to run your way, it runs your way: dedicated, on-premise, or air-gapped.

Talk to an engineer →

Predictable pricing.

A plan, then usage — metered per call against one organization balance, so every line traces back to a request. Here is the whole shape of it.

Go

$9/mo

$5 back as credit

Dev

$19/mo

$19 back as credit

Pro

$49/mo

$49 back as credit

Max

$99/mo

$99 back as credit

Above the entry plan, the fee is not a fee. Every dollar returns as credit you spend on inference, compute, or storage — so the plan buys you the software, and the usage is the only thing you are actually billed for.

Teams
$25 per person, minimum 2 seats. One org, one bill, SSO through Hanzo IAM.
Enterprise
Dedicated capacity, on-premise or air-gapped, and terms written for you. Talk to us.
Self-hosted
$0. The platform is open source — run the same image on your own cluster and bring your own provider keys.

Start with one bill.

Open the console and the whole platform is already provisioned against one organization. Or pull the image and run it yourself — it is the same artifact either way.

$ docker run -p 8080:8080 ghcr.io/hanzoai/cloud