Model Selection

Use GET /v1/models to choose a GonkaGate model ID for application code and keep model configurable. Use /models to browse the live catalog and /pricing when the question is billing rather than model selection.

Choose a current model ID

Start with a current model ID from GET /v1/models:

curl https://api.gonkagate.com/v1/models \
  -H "Authorization: Bearer gp-your-api-key"

{
  "object": "list",
  "data": [
    {
      "id": "qwen/qwen3-235b-a22b-instruct-2507-fp8",
      "object": "model",
      "created": 0,
      "owned_by": "gonka"
    }
  ]
}

Use one data[].id value as model in your next request.

Use the ID in a request

curl https://api.gonkagate.com/v1/chat/completions \
  -H "Authorization: Bearer gp-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen/qwen3-235b-a22b-instruct-2507-fp8",
    "messages": [
      { "role": "user", "content": "Summarize this document." }
    ]
  }'

Use any current data[].id from GET /v1/models. Keep model configurable so you can refresh it without changing application code.

Before you set a default model

• An API key that can call GET /v1/models
• One or two real prompts from your workload
• A config or environment variable for the default model value
• A note on whether the workload depends on tools, vision, structured outputs, or streaming

Common model ID issues

• 404 model_not_found means the ID is no longer valid for the request you sent. Refresh it from GET /v1/models.
• GET /v1/models can temporarily return 503 while the catalog is not ready. Retry with bounded backoff. Do not scrape /models or guess IDs manually.
• :online is not a canonical model ID returned by GET /v1/models. Start with the base data[].id. Append :online only as a request-time shorthand for web search on that call. For new GonkaGate integrations, prefer plugins: [{ "id": "web" }].
• /models is the human-readable catalog. It helps you browse, but GET /v1/models is the machine-readable source for application code.
• GET /v1/models returns minimal OpenAI-compatible model objects. Use /pricing for billing rules and /models for catalog context such as display names and capability notes.
• If your workload depends on tools, vision, or structured outputs, validate that capability with the model you plan to ship before rollout.
• A default model is not a permanent contract. Recheck it before releases, migrations, or fallback changes.

See also

• Quickstart: first request to send a working request with the model ID you selected here.
• GET /v1/models — List Available Models for the exact response schema.
• Live Models Catalog to browse current models before you copy an ID.
• Pricing if model choice is blocked by billing rules rather than API compatibility.