| Model | Input | Output | Cached input | Profile |
|---|---|---|---|---|
 doubao-seed-2.0-lite | 0.17 | 1.03 | 0.0343 | budgetfast |
 glm-5.3-flash | 0.23 | 0.80 | 0.0657 | budgetfast |
 deepseek-v4.1-flash Not for U.S. federal use | 0.57 | 2.29 | 0.0114 | balancedcheap cache |
doubao-seed-2.1-pro | 1.71 | 8.57 | 0.34 | flagship |
glm-5.3 | 2.29 | 8.00 | 0.57 | flagship |
deepseek-v4-pro Not for U.S. federal use | 2.57 | 7.71 | 0.0857 | flagship |
 kimi-k3 | 5.71 | 28.57 | 0.57 | premiumlong context |
Compliance notes
Under the FY2026 National Defense Authorization Act (§1532 and §6604) and the U.S. Department of Homeland Security directive issued in October 2026, U.S. federal civilian agencies, the Department of Defense, the Intelligence Community and their contractors may not procure or use DeepSeek technology — including through third-party services such as this one, and regardless of whether the model is self-hosted. Several states, including New York, Virginia, Tennessee and Texas, have imposed comparable restrictions on government devices and networks.
If you are a U.S. federal agency, a federal contractor, or a state or local government entity, do not use the DeepSeek models offered here. Other models in the catalogue are not covered by this restriction. This notice is provided for compliance convenience and is not legal advice; you remain responsible for your own regulatory position.
Every model in this catalogue is operated by a provider whose inference infrastructure is located in mainland China. When you call a model, the content of your request is transmitted to that provider and processed there, and the response is returned to you. Account and billing data are stored in the United States and the European Union.
We do not use your prompts or outputs to train models. If your workload cannot leave a particular jurisdiction, contact support@adzukiai.com before sending production traffic.
How billing works
Prepaid balance
You add credit up front. Each request deducts the cost of its input tokens, cached tokens and output tokens. When the balance reaches zero, requests stop.
Output costs more
Output tokens are priced several times higher than input, as with every provider. Cost is dominated by how much you generate, not how much you send.
Failed calls
A request that fails through our error is not charged. Credits are restored automatically; there is no ticket to open.
Frequently asked
Do I need a Chinese phone number or real-name verification?
No. You sign up with an email address and pay by card. There is no phone verification and no Chinese identity check. Nothing about the signup flow requires a mainland account.
How is usage billed?
Per token, in US dollars, against a prepaid balance. Input, cached input and output tokens are priced separately; output is the expensive part. There is no monthly minimum and no subscription. When the balance reaches zero, requests stop until you top up.
Do I have to change my code to use it?
No. The API speaks the OpenAI wire format, so an OpenAI SDK works after changing two values: the base URL and the API key. There is no proprietary SDK and no local proxy to run.
Where are requests processed?
Requests are routed to the model providers, whose infrastructure is in mainland China. That means data leaves your jurisdiction and is processed there. This is disclosed on every page that describes the service, and it is the one thing to check against your own compliance obligations before you build on it.
Are all of the models available everywhere?
Some DeepSeek models are restricted for US federal use. If you are a US federal entity or work under a federal contract, check the compliance notes above before depending on them.
Which models are available?
Seven models from four vendors: DeepSeek, Zhipu (GLM), Moonshot (Kimi) and ByteDance (Doubao). The table above lists the current lineup with per-million-token prices in USD.