For businesses
FastMetal for teams and high-volume use
FastMetal is an OpenAI-compatible gateway that gives you many LLMs behind one API key, billed in Japanese yen or US dollars (you choose the currency when you create your account) on a prepaid balance. This page collects what Japanese companies usually check before adopting it.
Why it fits Japanese development teams
Billed in yen or US dollars
Your account is charged in the currency you choose when you create it. On a yen account your monthly cost doesn't move with the exchange rate the way a dollar-denominated API does; a US-dollar account is there for teams that budget in dollars.
Qualified invoices
The operating company is a registered qualified invoice issuer in Japan (registration number T5010001015990), so invoices meet the requirements for input tax credit.
Prepaid, so spend can't overrun
You buy a balance and draw it down. Nothing is charged beyond it, so an unexpected large bill isn't structurally possible.
One key, many models
One OpenAI-compatible endpoint reaches models from several providers. No separate contract and billing relationship per provider.
What procurement and finance usually ask
Collected in a form you can take straight into an internal review.
| Item | Answer |
|---|---|
| Billing currency | Japanese yen (JPY) or US dollars (USD), chosen per account |
| Qualified invoice system | Registered qualified invoice issuer (T5010001015990) |
| Payment methods | Yen: credit card or Japanese bank transfer, both prepaid and both from ¥500. US dollars: credit card, prepaid, from $5. Bank transfer issues a qualified invoice immediately with a dedicated virtual account, payable for 14 days. There is no postpaid month-end invoicing |
| Spend cap | The balance you purchased is the cap. Nothing is billed beyond it |
| Training on your data | FastMetal does not use your data to train models |
| Where inference runs | Models whose id starts with japan- are hosted and run inside Japan (see the pricing page for the list). Other models run in their provider's region |
| Operating company | Kokusai Chotatsu Joho Co., Ltd. (Chiyoda-ku, Tokyo; established 1981) |
| API compatibility | OpenAI-compatible and Anthropic-compatible. Existing code works by changing the base URL |
Inference itself runs on each model provider's infrastructure, so their own policies govern what happens there. If you are working with sensitive data, talk to us first.
About payment — worth saying up front
At present we support prepayment only — there is no postpaid billing in arrears at month end. You purchase a balance and usage is drawn from it. You can pay by credit card or by bank transfer; choosing transfer issues a qualified invoice immediately, carrying a bank account dedicated to your organisation, and credit is added automatically once the transfer is confirmed, with nothing to notify us about. If that does not fit your requirements, please say so when you get in touch.
Using it across a team
- You can issue several API keys and set a separate balance (cap) on each, so budget can be split per team or per project.
- Usage is visible in the dashboard broken down by model and by day, so you can see where the spend went.
- Email lands when a balance runs low and again when it is exhausted — including for keys used only through the API.
Talking about volume
If your monthly usage is large, we're happy to discuss terms. Three things make that conversation concrete:
- Expected monthly token volume, or your current monthly spend
- Which models you plan to use — or just the use case, if that's not settled
- Preferred payment method and billing cycle
If you want to evaluate it technically first, the fastest route is to create an account and load a small balance. That's the same service, not a separate trial.