What is Seamhop?+
An AI model API gateway for developers: one key, one base_url for Claude, GPT, Gemini and more, routed across multiple providers.
Which clients are supported?+
Common clients and custom apps using compatible request formats.
How do I get a key?+
Register with your email in the console and verify it, then create a key (called a “token” in the console). Registration is open to everyone.
How are model names attributed?+
Model names belong to their respective owners.
How do I start?+
Register, get a key, then point your base_url to Seamhop.
Can I connect multiple models?+
Yes. Configure the model paths you want behind one integration surface.
How can I see usage and cost?+
The console dashboard summarises usage, and the usage log lists the model, input and output usage and cost of every call.
How do billing and top-ups work?+
Billing follows the input and output usage of each call; see the Models page for rates. Top-up options will be announced in the console when they open.
Can I get an invoice or pay by bank transfer?+
We can issue a commercial invoice or a receipt (mainland China VAT invoices are not included), and company transfers are supported. See the contact page for fees and the process, or email the billing address.
How can teams and companies work with you?+
Teams and companies are welcome; email us with your use case.
Which API formats are supported?+
OpenAI-compatible, Anthropic, OpenAI Responses and Gemini formats. The Models page lists the formats each model supports.
Where can I see prices?+
The Models page lists input and output rates per million tokens in both CNY and USD for each model.
Which models are available?+
Claude, GPT and Gemini, plus Chinese models including DeepSeek, GLM, Kimi, Qwen, MiniMax, Hunyuan and MiMo. See the Models page for the full list and the Models page for rates.
How do I write the endpoint and authentication?+
Point your base_url to https://api.seamhop.com/v1 and use the key you created in the console. The request example on the home page shows each format.
What happens when one path is unavailable?+
A model can have several paths. When one is unavailable, the request switches to a backup path automatically, with no change to your endpoint or key.
Why is the input usage of a short request higher than what I sent?+
For some models the upstream adds a system prompt, and those tokens count toward usage, so a short request can show more input usage than you sent. The console usage log is the source of truth for every call.
What is the “official list price” on the Models page?+
It comes from each vendor’s public page as of the date shown, for reference only; the vendor’s page prevails. Seamhop is not affiliated with these vendors.