Ai Model HubHelp Center Open workspace

Prepare Your First API Request

Copy the endpoint and model identifier from the documentation for the model you intend to call. The API hostname shown by Ai Model Hub differs from the website hostname. Confirm the model’s protocol, key permissions and price before running a request.

Find the current settings

Open More actions → API Docs, or select the current model’s documentation control. The current documentation lists:

Setting Documented value
API origin https://api.vv666.ai
Chat Completions endpoint POST /v1/chat/completions
Recommended authentication Authorization: Bearer YOUR_API_KEY
Example model identifier on the page gpt-6.2-astra

Use the exact identifier on your chosen model’s page. A display name and an API identifier can differ. The API Keys entry in Profile opens key management; keep keys in your application’s secret storage or server environment.

Minimal Chat Completions request

For a model documented as Chat Completions-compatible, the request has this shape:

curl --request POST 'https://api.vv666.ai/v1/chat/completions' \
  --header "Authorization: Bearer ${AI_MODEL_HUB_API_KEY}" \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "MODEL_ID_FROM_CURRENT_DOCS",
    "messages": [{"role": "user", "content": "Reply with a short greeting."}]
  }'

Set the environment variable securely and replace the model placeholder before running it. Sending the request can consume credits. Keep the first input short and use the current model documentation for any output limit you add.

For a client that appends /chat/completions to an OpenAI-style base URL, use https://api.vv666.ai/v1. For a client asking for a complete endpoint, use the full URL in the example. Check that the final path contains /v1 exactly once.

Check the task and its cost

The site documentation identifies the X-Task-Id response header as the platform task ID. It documents this lookup:

GET https://api.vv666.ai/v1/skills/task-status?task_id=TASK_ID

Keep the platform task ID with your request log. According to the documentation, cost is settled when is_final becomes true; the upstream response-body id is a different identifier. If settlement is still pending, query the existing task again instead of sending another generation request.

For an IP-restricted key, the documented error 403 with error.code = ip_not_allowed means the request came from an address outside that key’s allowlist. Review the intended application’s outbound address and key settings.

Source: current site API documentation. Next: Understand credits.

Last updated

Need help with a specific task?

Open the workspace, then choose More actions and Contact support. Keep the task ID, model and error message ready.

Open workspace