Regional
Suppliers processing in our Regional zone. The default for every new account: the strongest data control, with a growing list of models.
OpenAI-compatible API with data-zone control
One key.
Every open model.
DeepSeek, Qwen, GLM, Kimi, FLUX and more behind a single endpoint. You decide where your requests are processed, per account and per key, and pay only for the tokens you use.
- Regional or External processing
- Automatic failover
- Invoices for business
- Company verification and credit approval
Live cost demo
Know the cost before you send a request
Pick a model and the size of a typical request. Numbers come from the live price list below.
Your prompt, chat history and any documents you send.
What the model writes back. A token is about ¾ of an English word.
Models & pricing
B2B pilot · monthly invoicing
Indicative catalogue prices in USD. Confirm your business terms with HQ before API approval.
| Model | Type | Where it runs | Context | Input | Output |
|---|---|---|---|---|---|
| Loading prices… | |||||
Data zones
Decide where your requests are processed
Every supplier we route to sits in a data zone. Your account and each API key say which zones may handle your data, and we never move a request into a zone you have not allowed, not even when a supplier is down.
External
Suppliers outside it, such as model makers' own APIs, including in China. The widest choice and the lowest prices. Used only after you opt in, and you can limit single keys to Regional.
Private
For enterprise customers: models served from your own cloud account or data centre, reachable only by your keys. Talk to us.
Each response carries x-data-zone and x-processing-region headers, and your usage log shows the zone and region of every request.
Drop-in compatible
Change two lines. Keep the rest of your code.
Works with the official OpenAI SDKs, LangChain, LlamaIndex, the Vercel AI SDK and any tool that lets you set a base URL.
- Streaming, tool calls and JSON mode
- Vision models accept images in messages
- Embeddings, rerank, images and speech with the same key
Why AI Master
Built for teams that ship
Always on
Popular models are served by more than one supplier. If one fails or runs out of capacity, your request moves to the next one in the zones you allow, without you changing a line of code.
Available credit allowance
HQ limits apply to every key. A key can be more restrictive. No selected models blocks every model.
Limits for every key
Give each app or teammate its own key, with a monthly spending cap and a list of allowed models.
B2B pilot · monthly invoicing
Arrange month-end settlement with your account manager. Online checkout is disabled for the B2B pilot.
Receipts finance accepts
Every payment gets a numbered receipt with your company name, address and tax ID, ready to print or save as PDF.
Usage down to the request
Tokens, cost and latency for every call, filterable by key and model. Export to CSV whenever you need it.
Your prompts stay yours
We record usage metadata for billing. We don't store your prompts or the model's answers.
Stable model names
Use aimaster-fast or aimaster-pro and we keep them pointed at the best current model, so a model retiring never breaks your app.
Questions
Does it work with the OpenAI SDK?
Yes. Set the base URL to https://www.aimaster.cloud/v1 and use your AI Master key as the API key. Model IDs are in the price list above and at GET /v1/models.
How does billing work?
Business pilot: register your company, get HQ approval, and pay monthly for usage within agreed limits.
Which payment methods do you accept?
Arrange month-end settlement with your account manager. Online checkout is disabled for the B2B pilot.
Can I get an invoice for my company?
Yes. Choose a business account and add your company name, address and tax ID in Settings. Every payment then gets a numbered receipt you can print or save as PDF.
Where is my data processed?
That is up to you. Each supplier is in the Regional or the External zone, and requests only go to zones your account allows; you can also limit single API keys to Regional. New accounts choose at sign-up. Every response and every line of your usage log shows the zone and region that handled it.
What happens if a model provider goes down?
Where a model has more than one supplier in your allowed zones, we retry the next one automatically. You are only charged for the request that succeeds.
Do you store my prompts?
No. We record request metadata (time, model, token counts, cost and latency) for billing and support. Prompts and outputs pass through to the model and are not stored by us. See the privacy policy.
Can I get my money back?
See the refund policy for when unused credit can be refunded.
Apply for a business account
Business pilot: register your company, get HQ approval, and pay monthly for usage within agreed limits.
Apply for a business account