One request.
A flexible foundation.

Opaline’s proposed inference interface uses a model identifier, message history, and generation parameters. The playground creates this payload locally so developers can explore its structure.

The request structure

{
  "model": "<your-model-id>",
  "messages": [
    { "role": "user", "content": "Hello, Opaline." }
  ],
  "temperature": 0.7,
  "stream": true
}

Model-specific capabilities

Streaming, vision, tool interaction, and structured generation depend on the chosen model and provider. Do not assume every model supports every request parameter.

Before production integration

• Configure an authenticated inference service and a supported model ID.
• Keep credentials on a server and apply project-level access controls.
• Validate input, handle provider errors, and enforce spending limits.
• Measure actual usage from provider responses.