Core Engine
Chat Completion API Layer
Send a sequence of role/content messages to an enabled chat model. The response includes the generated content, provider, model, token usage, and latency.
Endpoint Structure
POST
https://aiservice.prabisha.com/chat
Payload Paradigm Spec
{
"messages": [
{
"role": "user",
"content": "Explain core caching patterns."
}
],
"model": "gemini-2.5-flash",
"temperature": 0.7,
"maxTokens": 1000,
"preferredProvider": "OPENAI"
}
Request Field Configurations
| Parameter | Type | Required | Description Rule |
|---|---|---|---|
| messages | Array | True | Required. Each item has a role and content; common roles are user and assistant. |
| model | String | False | Optional enabled model ID. If omitted, the router selects an active default text model. |
| temperature | Number | False | Probability spread variance metrics control boundaries [0-2]. Default 0.7 |
| preferredProvider | String | False | Target strict routing bypass parameters: OPENAI, GEMINI, etc. |
| maxTokens | Number | Optional positive output token limit. Provider support may vary. |
Unified Output Object Shape
{
"success": true,
"data": {
"content": "Systems logic returned successfully.",
"model": "gpt-4o",
"providerName": "GEMINI",
"providerId": "provider-id",
"providerModelId": "provider-model-id",
"fallbackChain": ["GEMINI"],
"usage": {
"promptTokens": 15,
"completionTokens": 120,
"totalTokens": 135
},
"latency": 342
}
}
API Request Sandbox Terminal
Send a live request to the production chat endpoint.