Frequently asked questions

Tovano is one API and one bill for AI models: image, video, audio and text. Your team uses one API key, one prepaid balance in US dollars and one monthly invoice, and every call is charged per unit at a published price. It is built for companies that run AI features inside their own products at volume.

Product and engineering teams that use several AI models in production and serve many end customers of their own: ad and creative tools, commerce platforms, media companies, support products and agencies. Most of our customers spend tens of thousands of dollars a month on models and need that spend to be predictable, traceable per customer and easy for finance to approve.

The current catalogue, with unit, price and input fields for every model, is on the models page and at GET /v1/models. It covers image, video, speech and text models and agent jobs that chain several models. New models appear in the same API with their own unit and price, so your integration does not change.

A task that runs several models in a row on our side and returns one result. agent/ad-kit-1, for example, writes ad copy with a text model and then makes a matching image. Each step is listed on the task with its own charge, and if any step fails the whole job is refunded. See Run an agent job.

Where a model is offered by more than one provider, we route the task to another provider at the same price. Where it is not, the task fails with 502 provider_error and is not charged. Current availability is on the status page.

Each model has one unit and one price per unit: per image, per second of video, per 1,000 characters of speech, or per million input and output tokens for text. You pay that price for what the model delivered. There are no seat fees and no platform fees. Prices are on the pricing page.

Prepaid credits let us publish one price for everyone and start every request without a credit check. They also give your finance team a hard ceiling: spend can never pass the balance you funded, and you can add a monthly spend limit on top. Credits are bought in US dollars and do not expire.

Yes. Larger top-ups add bonus credits: 2% from $1,000, 5% from $2,500 and 8% from $10,000 per top-up. The bonus is added as a separate line in your ledger. For commitments beyond that, talk to sales@tovano.ai.

No. When a task starts, its estimated cost is reserved on your balance. When it completes, you are charged the real amount. When it fails, the whole reservation is released and the task shows charged_micro: 0. This holds for errors on our side and on the provider's.

By card through our payment processor, as a one-off top-up or as an automatic top-up when the balance drops below a threshold you choose. Every top-up gets a receipt. For larger accounts that need to pay by bank transfer against an invoice, write to sales@tovano.ai.

One invoice per month for the organisation, with usage per model (units, tasks and amount), top-ups, bonus credits and the closing balance. Add your company name, address and VAT ID under Billing and they appear on every invoice. Invoices and receipts are available as PDF.

Unused credits from a top-up can be refunded within 14 days of that top-up. Bonus credits have no cash value. When you close your account, unused paid credits are refunded under the same terms. Details are in the terms.

No. Prices are in US dollars, per unit, excluding tax. We add VAT or sales tax where the law requires it.

Most teams make their first call within an hour and move production traffic within days. Every model uses the same key, headers and error format, and there is a TypeScript SDK, an OpenAPI description and a Postman collection to start from.

Yes. Text models use the widely used chat completions format at /v1/chat/completions, including streaming. In most cases you change the base URL to https://api.tovano.ai/v1, the API key and the model name. See Stream chat replies.

Pass a webhook_url with the task, or set a default per project, and we POST the finished task to you with a signature you can check. Failed deliveries are retried three times. Polling GET /v1/tasks/{id} works too. See Receive results with webhooks.

Yes. Send your own customer ID with each call in the customer field or the X-Tovano-Customer header, then call GET /v1/usage?group_by=customer. The same view is in the dashboard with a CSV export. See Track spend per end customer.

Use projects. Each project has its own API keys, webhook settings and usage, while balance and invoices stay with the organisation. Usage can be grouped by project or by API key, so each team can see its own spend.

Limits apply per API key and are returned on every response in X-RateLimit-Limit and X-RateLimit-Remaining. Tasks above your concurrency limit wait as queued instead of being refused. For production volume we size limits to your traffic: tell sales@tovano.ai your expected peak before a launch.

Set a monthly spend limit for the organisation under Billing. Requests that would pass it are refused with 402 spend_limit_reached. Add a low-balance alert and, if you want production never to stop, auto top-up. See Set spend limits and low-balance alerts.

Send an Idempotency-Key header when you create a task. A retry with the same key and body returns the first result instead of running the task again. Keys are kept for 24 hours. See Handle errors and retry safely.

99.9% monthly availability of the API, measured on the status page. Planned maintenance is announced at least 48 hours ahead.

No. We do not use your inputs or outputs to train models. We process them only to run your requests. You keep all rights to your inputs and outputs.

Output files are reachable through signed URLs that work for 24 hours, and are deleted after 30 days. Download what you want to keep to your own storage.

The API, database and stored outputs are hosted in the EU, in Germany. Some model providers process data outside the EU under the EU Standard Contractual Clauses. Each one is listed with its location on the subprocessors page, and our DPA covers the rest.

A key is shown once when you create it and stored only as a SHA-256 hash and a short prefix, so nobody at Tovano can read it back. Revoke a key in the dashboard and it stops working at once. Every key change is recorded in the activity log. More on the security page.

Yes. Invite colleagues to your organisation as owners or members. Owners manage billing, limits and members; members work with projects, keys and usage. Everyone signs in with an email link, Google or GitHub, and every sign-in is in the activity log.