Skip to content

One AI stack.
Any model. Any cloud.

AI teammates do the work. The best model takes every task. It all runs on the right cloud.

Describe your businessSee the stack
The ZB Stack workspace, headed “The workspace for any model, any cloud.” and “What are you building?”: a box with a candle store described in it, the five models that will share the build (spec on Fable, plan on GPT, code on DeepSeek, tests on Sonnet, review on Opus), and below it the AI teams by industry and the twelve models grouped by the work they do.
Works with
  • Claude
  • GPT
  • Gemini
  • DeepSeek
  • Grok
  • Mistral
  • Qwen
  • GLM
  • Llama

01 / AI teams

Put an AI team to work.

Give your AI team a goal. The Lead makes the plan and hands out the work: product, design, code, marketing, sales, operations, service. Finished work lands in one workspace. And your AI team already knows how businesses like yours sell, ship and serve customers.

Work beside them in shared channels with your cofounder and contractors.

Meet the AI teams
Illustrative workspaceAI teams that know your industry

What one AI team builds for an online store

Storefront
  • Storefront
  • Shopping agent
  • Checkout and payments
Growth
  • Marketing platform
  • Email and SMS
  • Social media
  • Search and SEO
  • Ads and retargeting
  • Reviews and loyalty
  • CRM
Operations
  • Orders
  • Inventory
  • Shipping
  • Returns
  • Tax and accounting
  • Customer service
Run
  • Analytics
  • Hosting on ZB Cloud
  • Monitoring and backups
One shared plan
  • Lead
  • Product
  • Design
  • Coding
  • Marketing
  • Sales
  • Operations
  • Service

Your AI team plans the work. The Model Router chooses the model for each task. ZB Cloud runs it.

02 / Model Router

The right model for every task.

Each task goes to the model that does it best, with everything done so far. The result is checked before the next model takes over. A failed check goes to another model. Spec from one, code from another, review from a third. Nothing re-explained. Or pick the model yourself.

GPT drafting, planningDeepSeek summariesQwen extractionFable specsGemini imagesGLM translationAstra hard planningLlama cleanupSonnet testingOpus reviewMistral taggingGrok researchCheckedFinished workdone and shippedproduct spectestingextractcode reviewsummarizeresearchlong buildtranslatedraft emaildata cleanupimageplanningtaggingAI teammatesnew tasksModel Router

Tasks from the AI teammates pass through the Model Router, which routes each one along a dotted line to one of twelve models: GPT for drafting and planning, DeepSeek for summaries, Qwen for extraction, Fable for specs, Gemini for images, GLM for translation, Astra for hard planning, Llama for cleanup, Sonnet for testing, Opus for review, Mistral for tagging and Grok for research. Each result passes a check gate marked Checked before it reaches a box marked Finished work, done and shipped. One long build fails its check at Sonnet and is moved to Opus, where it passes. Illustrative routing, not a benchmark.

Models from supported providers
  • OpenAI
  • Anthropic
  • Google
  • DeepSeek
  • xAI
  • Mistral
  • + more

The models

300+ AI model endpoints across 14 leading model providers.

Including OpenAI, Anthropic, Google Gemini, xAI, DeepSeek, Qwen, GLM, Kimi, MiniMax, Mistral, NVIDIA, Cohere, Perplexity and more. Example model choices today:

  • SpecsFable
  • Drafting, planningGPT
  • Code, summariesDeepSeek
  • TestingSonnet
  • ReviewOpus
  • ResearchGrok
  • ImagesGemini
  • TranslationGLM
  • ExtractionQwen
  • CleanupLlama
  • TaggingMistral
  • Hard planningAstra

When a model does a job better or cheaper, the board changes. You can see which model ran each task and what it cost.

03 / ZB Cloud

Compute where you want.

Our cloud runs on every cloud. Each job goes to the cloud that should run it, chosen by price, capacity and location. When one of those changes, the work moves.

Sensitive data stays inside the border. A nightly job runs wherever is cheapest tonight. You see where every job runs and what it costs. Zero DevOps: the system operates itself.

Explore ZB Cloud
Illustrative cloud selection

One example, priced three ways

The same build, three prices.

Northwick Candle Co. is our scripted example: a storefront, a shopping agent and eighteen systems behind them. 3,559 AI tasks. Everything on Fable: $720.00. Everything on Opus: $290.97. Routed by the Model Router: $29.10, fee included.

The sample build

Northwick Candle Co.

A storefront, a shopping agent and eighteen systems behind them.

  • Storefront
  • Payments
  • Orders
  • Marketing
  • Customer support
  • 20 systems
  • 3,559 tasks
  • 12 models
Fable, one model for everything$720.00
Opus, one model for everything$290.97
ZB Stack, routed by the Model Router, with the 5% fee$29.10
90% less than Opus · 96% less than Fable
The record

Every task. Its model. Its cost.

Open any line and see why that model got that job.

Explore the example workflow
TaskModelCost
Write the store spec: 20 systems, 40 products, two currenciesFable$1.07
Spec the storefront: home, collections, product pages, cartFable$0.70
Spec checkout and payments in USD and CADFable$0.55
Review the whole store before it goes liveOpus$0.47
Write the abandoned-cart email, four filesDeepSeek$0.04

Scripted beta example, not a customer result. One model = Claude Fable 5.1 or Claude Opus 5.5, at list price on 10 October 2026. Image tasks are priced on Gemini in both columns. Row figures are before the 5% fee; the $29.10 total includes it. Hosting is separate.

Pricing

Model usage at cost, plus 5%.

  • AI teammates included.
  • No monthly subscription.
  • No charge per AI teammate.
  • Hosting is charged separately, at provider cost.
Describe your business
A simple example
Model usage
$10.00
ZB Stack fee · 5%
$0.50
Total
$10.50

Put money in. Spend it on any model. Nothing is charged until a task runs.

Who it is for

Built for four kinds of builder.

  • Governments

    Build sovereign AI.

    Sensitive work stays inside the border. Everything else runs on the best cloud for the job.

    The sovereign AI offering
  • Startups

    Build and run a company with AI.

    Bring an idea, a prototype or a live business. Put AI teammates on the work ahead.

    For startups
  • Enterprises

    Switch models the day a better one ships.

    Every model and every cloud through one stack, and the cost of every task in the record.

    Write to us
  • Developers

    Every model through one door.

    Route work by cost, speed, quality and your rules.

    How it works

Questions before you start

Read the full FAQ
What does it cost?

Model usage at the provider's price, plus 5%. That is the only fee. Ten dollars of model usage is fifty cents to ZB Stack, $10.50 in all. Hosting is separate, at provider cost. There is no monthly ZB Stack subscription.

What is ZB Cloud?

Our cloud runs on every cloud. It puts each workload where price, capacity and your location rules say it should go.

Do I need a ChatGPT or Claude subscription?

No. ZB Stack model usage is billed separately from those subscriptions. If you have an API key with a provider, bring it; your usage on that provider is billed to you directly, with the same 5%.

Which models?

Twelve on the board today, from nine companies, chosen from 300+ AI model endpoints across 14 leading model providers. The board changes when a model does a job better or cheaper.

Which clouds?

The big global clouds, sovereign clouds inside a country’s border, specialist GPU clouds and decentralized networks of independent machines. A workload that must stay inside a border stays there.

Can I choose the model myself?

Yes, on any task, or let the Router choose. The same for the cloud a workload runs on.

Is my balance a subscription?

No. You add funds and use them as tasks run. Nothing is charged until a task runs.

Who owns what gets built?

You do. Your code, your data, your systems and your business stay yours.

Your next move

Build, operate, and scale AI your way.

Tell us what you’re building.

Describe your business

Enterprise or government project? Write to us.