Request Parameters

Share

KiteRouter accepts the OpenAI-compatible chat completion payload and forwards it to the selected upstream OpenAI-compatible provider route.

Parameters

ParameterTypeRequiredDescription
modelstringKiteRouter model ID, for example gpt-4o-mini. This can map to one or more provider models.
messagesarrayOrdered chat messages with role and content.
streambooleanIf true, returns Server-Sent Events. Default: false.
temperaturenumberSampling temperature. Forwarded as-is to the upstream provider.
max_tokensintegerMaximum output tokens when supported by the upstream provider.
top_pnumberNucleus sampling threshold. Forwarded as-is.
stopstring | arrayStop sequence(s). Forwarded as-is.
toolsarrayOpenAI-compatible tool definitions for models that support tool calling.
tool_choicestring | objectTool selection behavior. Forwarded as-is.
response_formatobjectJSON mode / structured output options when supported by the model.

Message Format

json
1{
2  "role": "user",
3  "content": "Your message here"
4}

Roles

RoleDescription
systemSets assistant behavior. KiteRouter may inject an admin-configured system prompt before forwarding.
userUser input.
assistantPrevious assistant response for conversation context.
toolTool result messages when using OpenAI-compatible tool calling.

Info: KiteRouter does not convert request formats. Use OpenAI-compatible payloads for /api/v1/chat/completions.

Ready to start building?

Get your API key and make your first request in under a minute.