Perf Chat icon

Perf Chat

Send messages to any LLM with intelligent model routing. One node, any model — Perf automatically selects the best model for each request.

Perf Chat

AI-generated

Summary

Send a chat message to an LLM with intelligent model routing, supporting simple text, multi‑turn messages, or raw JSON input.

Inputs

  • inputMode — How to provide the input messages (simple, messages, or json).
  • message (required) — The single user message when inputMode is simple.
  • messages — Array of role‑content messages for conversation context.
  • messagesJson — Messages array in OpenAI JSON format as a string or parsed object.
  • model — Target model to use; auto selects the best model if set to 'auto'.
  • systemPrompt — Optional system prompt prepended to the conversation.
  • options.maxCostPerCall — Maximum cost in USD for the call; 0 means no limit.
  • options.responseFormat — Response format (text or json).
  • options.jsonSchema — JSON Schema to enforce on the response when responseFormat is json.
  • options.includeMetadata — Whether to include routing metadata (model, cost, latency) in the output.

Output shape

array of objects representing chat responses

Each item always contains content, role, model, and finish_reason; may also include json_content, perf metadata, and usage fields depending on configuration.

Examples

Example 1: Send a single user message with intelligent routing

inputMode=simple, message="Hello world", model=auto

Example 2: Send a multi‑turn conversation using JSON format

inputMode=json, messagesJson='[{"role":"user","content":"Explain quantum computing"}]', model=gpt-4o

Example 3: Enforce a JSON schema on the response

inputMode=simple, message="List tools", responseFormat=json, jsonSchema='{"type":"object","properties":{"tools":{"type":"array"}}}'

Links

Discussion