Petals MemoryInstall

Summary

Send a non-streaming chat-completion request to the selected Petals assistant and return its response. Build the conversation from message fields or provide an OpenAI messages array as JSON; optional completion settings are passed through to the endpoint.

Inputs

• Assistant Name or ID (required) — Select an assistant or supply an ID. Use assistantId@modelId to pin a model.
• Messages Input — Choose Fields (default) to enter messages individually, or JSON to provide an OpenAI messages array.
• Messages — In Fields mode, add messages with a role and text content; name and tool call ID are optional. Tool call ID is sent as tool_call_id.
• Messages (JSON) — In JSON mode, provide an OpenAI messages array, including content parts, tool calls, or tool results.
• Additional Fields — Optional completion parameters include frequency and presence penalties, logit bias, max tokens, metadata, modalities, number of choices, parallel tool calls, response format, seed, stop sequences, temperature, tool choice, tools, top P, and user identifier.

Output shape

One or more items containing the chat-completion response object.

Each returned response row is emitted as an item paired with its input item. The number of rows depends on the endpoint response; the Number of Choices setting requests how many answers to generate. Streaming is not supported.

Links

• n8n community nodes installation guide