API essentials

This page covers documented examples, not a complete upstream API specification. Client feature support depends on its protocol and the model/key group.

Protocol	Base URL	Example endpoint
OpenAI-compatible	https://api.nekocloud.vip/v1	POST /chat/completions or POST /responses
Anthropic-compatible	https://api.nekocloud.vip	Use the client SDK's Anthropic API path
Gemini-compatible	https://api.nekocloud.vip/v1beta	Use the client SDK's Gemini API path

Authentication and first request

Create an API key on the main site. Send it in Authorization: Bearer <your-key> for OpenAI-compatible requests. Do not publish it in code, URLs, screenshots, or issue reports.

curl https://api.nekocloud.vip/v1/chat/completions \
  -H "Authorization: Bearer sk-your-key" \
  -H "Content-Type: application/json" \
  -d '{"model":"your-available-model","messages":[{"role":"user","content":"Hello"}]}'

Replace the example model with one accessible to your key's group. For Codex, use the Responses setup instead of assuming chat completions work for every model. See Model Plaza for displayed prices and units; Usage on the main site is the per-request billing record.

Streaming and errors

For a client that supports streaming chat completions, set "stream": true; consume the event stream incrementally rather than buffering it as JSON. Clients should handle interrupted streams without assuming the entire answer was delivered.

Status	First check
401	Is the Authorization header present and the key valid?
403	Balance, account status, or applicable safety policy; inspect the response error code
429	Reduce concurrency, respect any retry guidance, then back off
Model unavailable	Exact model name and key-group access

Do not blindly retry non-idempotent requests: they may already have reached the upstream. For persistent failures, record the time, model, response code and request ID (if present), but redact keys and user input before contacting support.