FAQ — Model API

General

What is mass.gogpu.cn?

It is the HTTP API base URL for Gogpu Cloud hosted LLMs — not the console website. Configure https://mass.gogpu.cn/v1 (OpenAI-style) or /anthropic in your code or AI tools.

Is Model API the same as “model marketplace” in console?

Model API = call models over HTTP with an API key. GPU model marketplace (Chinese docs) = deploy open-source models on your own GPU VM. Different products and billing.

Do I need a GPU VM to use Model API?

No. Model API runs on Gogpu Cloud infrastructure. You only need a console account, billing setup, and an API key.

Keys & auth

Where do I create keys?

console.gogpu.cnopen in new window → Model API → API KEY.

I lost my key

Keys are shown once. Revoke the old key and create a new one.

Models & requests

How do I get the correct model name?

Call GET https://mass.gogpu.cn/v1/models with your key, or copy the API model id from console. Names are case-sensitive.

Which API style should I use?

If the model doc says OpenAI-compatible, use /v1/chat/completions. If Anthropic-compatible, use /anthropic/v1/messages. See API Guide.

Billing errors

Balance insufficient

Top up for pay-as-you-go, or switch to a key bound to a valid token package.

Package exhausted

Purchase a new package or use pay-as-you-go.

OpenClaw & agents

OpenClaw on GPU VMs can use local models (Ollama) or route to Model API. Full setup is in the Chinese OpenClaw guide. Set baseUrl to https://mass.gogpu.cn/v1 in openclaw.json.