Partners and OEM
Build FastMetal into your own service
For companies who want to offer AI features to their own customers, we expose a provisioning API that creates accounts, issues API keys and allocates budget programmatically — so per-customer keys and spend limits stay under your system's control.
Who this is for
Hosting providers
Adding AI features to an existing control panel. Account creation and key issuance happen inside the signup flow you already have.
System integrators and agencies
Building AI into what you deliver per client. Keys and budgets split per engagement, so usage attribution and cost pass-through stay simple.
SaaS and app builders
Shipping AI as a feature of your own product. Issue a key per end user with its own cap, so one heavy user can't drain the whole budget.
What you can do programmatically
All of the below are backend calls. None of it requires anyone to click through an admin UI.
| Operation | What it does |
|---|---|
| Create an account | Provisions a customer account and an API key in one call. Idempotency keys are supported, so a lost response never results in a duplicate |
| Allocate budget | Sets a spend cap per key. Requests are rejected once the cap is reached, so there is no overage to reconcile |
| Look up an account | Retrieves account metadata by exact email or id |
| Validate a key and read balance | Confirms a key you hold is valid and returns its cap, spend and remaining balance — ready to render in your own dashboard |
The full specification goes to companies who get in touch.
Technical ground rules
- The endpoint is OpenAI-compatible. Existing OpenAI-targeted code works by changing the base URL.
- One key reaches models from several providers. You don't need a contract with each of them.
- Billing is in Japanese yen. The operating company is a registered qualified invoice issuer in Japan (T5010001015990).
- Prepaid, so usage cannot exceed the budget you allocated.
About commercial terms
We don't publish them — they're set per partner. What fits depends on how many customers you expect and at what scale, whether billing sits with you or with us, and how support is divided. Rather than fit you to a fixed template, we'd rather hear how you're set up and design from there.
An early-stage idea is fine. We're happy to start with whether the thing you have in mind is possible at all.
What helps us answer quickly
Three things let us tell you fast whether this is workable:
- What your business does, and which customers you want to reach with AI features
- Roughly how many customers, and at what usage scale
- Whether billing and support should sit with you or with us
Frequently asked questions
- Can FastMetal be embedded in our own service or resold?
- We expose a provisioning API that creates accounts, issues API keys and allocates budget programmatically, so per-customer keys and spend limits stay under your system’s control. The full specification goes to companies who get in touch.
- Can keys and spend limits be separated per customer?
- Yes. Keys are issued per customer with a spend cap on each. Requests are rejected once a cap is reached, so there is no overage. A key’s balance and spend can be read back through the API and rendered in your own dashboard.
- Are partner pricing and terms published?
- No. Terms are set per partner, because what fits depends on the number and scale of customers, whether billing sits with you or with us, and how support is divided. Get in touch and we will work it out.
- Is there a minimum commitment?
- None is defined. Early-stage conversations are welcome, including simply whether what you have in mind is possible.