Bank account
$ 1,986,315.42
An account in your business name, to receive and send money by bank transfer.
- Local and international transfers
- Sub-accounts for teams, projects and agents
One business account for your money and your AI. Hold cash and stablecoins, spend by card, and call leading text, image and video models. Every call and every agent payment is metered and paid from your balance in real time. No new vendor contracts. No separate AI bills.
Text
- Claude
- GPT
- Gemini
- DeepSeek
- Qwen
- GLM
Video and image
- Veo
- Kling
- Seedance
- Wan
- Runway
Agents
- MCP
- x402
- AP2
One balance for cash, stablecoins and cards.
iTabb starts as a business account. Money arrives by bank transfer or as stablecoins, and leaves by transfer, by card or as a model call. It is all the same balance.
Stablecoins
500,000.00 USDC
Hold stablecoins next to cash, and convert either way when you need to.
- Receive and pay in stablecoins
- Settles agent payments over x402
Cards
14 active
Virtual and physical cards for people and for agents, each with its own limit.
- One card per team, project or agent
- Freeze or cancel at any time
Together $2,486,315.42, the balance that every model call and agent payment draws on.
AI spend lands on the same statement as everything else.
Every charge is a line item showing the model or payee, the team or agent, and the usage: tokens, seconds of video, or an agent's payment. Finance no longer logs in to each vendor's dashboard to export invoices, and there are no foreign card charges to reconcile.
Business account statementAccount ending 8821Nov 11–12, 2026
| Date | Type | Counterparty or model | Usage | Amount (USD) | Balance (USD) |
|---|
0 AI and agent charges this period, 0 tokensTotal charged $0.00
No procurement cycle for every new model.
iTabb already holds the contracts with model providers and settles with them for you. Turn on AI access and every connected model is ready.
Contracting each provider yourself
- Vendor assessment and security review, one provider at a time
- Negotiate contracts and data processing terms, then wait for legal
- Get a corporate card approved, or prepay the vendor
- A separate invoice from every provider, reconciled by hand
Usually takes weeks, and starts over with every new provider.
Calling models inside iTabb
- Turn on AI access in your dashboardIf you already hold an iTabb account, there is nothing new to sign.
- Create API keys and assign them to teamsEach key gets its own budget and list of models.
- Every call is paid from your balanceAll model spend is combined into one monthly invoice.
Live the same day.
The right model for each task, chosen for you.
Most work does not need the most expensive model, and some work should never be sent to a cheap one. Set the model to auto and iTabb reads each task, picks the model that fits, and shows what it chose and why.
Routed to
DeepSeek
Short inputs and a fixed set of labels. A light model matches the flagship on this kind of work.
- Short inputs
- High volume
- Fixed labels
- This route
- $36.00
- Flagship only
- $1,200.00
- Saved
- 97%
"model": "auto" returns "routed_to": "deepseek"More than a gateway.
Access to models is the easy part. These are the things that make the bill smaller and the service dependable.
Automatic failover
When a provider slows down or goes offline, calls move to an equivalent model. Your teams and agents keep working.
Prompt caching
Context you send again and again is cached and billed at a fraction of the price. The saving is shown on each call.
Batch pricing
Work that can wait until morning runs overnight at a lower rate, with the same models and the same statement.
Test on your own data
Upload a sample of real work and compare models on quality, cost and speed before you move a workload.
Data redaction
Card numbers, ID numbers and other personal data can be masked before a prompt leaves your account.
Savings report
Each month, see which workloads could move to a cheaper model without losing quality, and what that would save.
Give every agent a budget, not a card.
Agents call models and pay for the tools, data and suppliers they need from a sub-account you fund. Each one works inside a mandate you set: how much it can spend, who it can pay, and when a person has to approve.
- Plan the shipmentCalls Claude Sonnet, 18,420 tokens−$0.09Queued
- Fetch live freight quotesPays Freight rates API over x402, 6 requests−$0.12Queued
- Render a route briefing for the customerCalls Veo, 8 seconds of video−$3.20Queued
- Pay the carrier depositPays Harbourline Freight Co., 1 invoice−$1,860.00Queued
Run the agent against the demo account. Small charges settle on their own; the deposit is over the limit and waits for you.
Agents call
Agents use the same API as your teams, or connect over MCP to call models, check their remaining budget and request payments as tools.
Agents pay
Pay-per-request APIs settle over x402 in the moment. Suppliers and invoices are paid by bank transfer, card or stablecoin, each with a signed mandate the payee can verify.
People approve
Anything over the limit, or to a payee that is not on the list, waits for a named approver. Pause or revoke an agent at any time.
Pay only for what you use.
No monthly fee, no minimum spend, nothing prepaid to a vendor. Your funds stay in your iTabb account until a call actually uses them.
- Text is metered per token, video per second and images per image
- Charged to your balance as soon as each call completes
- Calls pause when funds run out, so you never carry a debt
- Agents spend from the same balance, inside the limits you set
- Spend is grouped by team, project, agent and model automatically
Finance sees every dollar of AI spend.
Budgets, permissions and usage live in your iTabb dashboard, managed the same way you already approve payments.
Team budgets
Give each team or project its own budget. When it runs out, calls stop without touching anyone else's funds.
Spend alerts
Notify the owner and finance when the balance or monthly spend crosses a threshold you set.
Per-key permissions
Each API key is tied to a team, project or agent and a list of allowed models. Revoke it at any time.
Live usage
See tokens, video seconds and agent payments by model, team and project, and export it to your finance system.
One invoice
Spend across every model and every agent is combined into a single monthly invoice.
No training on your data
Prompts and outputs are never used to train models. You decide how long logs are kept.
One key for every model and every agent.
The API follows the request format of the major SDKs, so migrating means changing one URL. Every response includes the amount charged and your balance afterwards, so an agent always knows what it has left.
- Endpoint
api.itabb.ai/v1 - Text at
/chat/completions, video at/videos, agent payments at/agent/payments - Agents can connect over MCP at
mcp.itabb.ai - One key can call every model, or be limited to a few
- Send
"model": "auto"and the response names the model chosen inrouted_to - Returns
402when funds run low, naming the budget that is short
curl https://api.itabb.ai/v1/chat/completions \ -H "Authorization: Bearer $ITABB_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "claude-sonnet", "messages": [ { "role": "user", "content": "List the payment terms in this contract" } ] }'
{
"id": "chatcmpl_7Hq2xV",
"model": "claude-sonnet",
"usage": {
"input_tokens": 3120,
"output_tokens": 486
},
"billing": {
"charged": "0.02",
"currency": "USD",
"balance_after": "2486315.42",
"sub_account": "R&D"
}
}Send the request to run it against the demo account. The charge comes out of the balance above.
Put your balance
to work.
One account for cash, stablecoins, cards and AI. Book a demo and we will set it up around how your team works.