SiliconFlowInstall

Summary

Sends the supplied messages to a SiliconFlow chat completion model and returns the API response for each input item. Uses the required SiliconFlow API credentials and sends the request to /chat/completions with streaming disabled.

Inputs

• Model (required) — SiliconFlow model ID; defaults to Qwen/Qwen2.5-7B-Instruct.
• Messages (required) — OpenAI-format messages array, which can use expressions to reference incoming data.
• Temperature — Optional sampling temperature from 0 to 2; UI default is 0.7.
• Max Tokens — Optional maximum number of generated tokens; UI default is 512.
• Top P — Optional top-p sampling value from 0 to 1; UI default is 0.9.
• Stream — UI option defaults to false, but the implementation always sends stream: false; enabling it does not enable streaming.

Output shape

One item per input item, containing the SiliconFlow chat completion API response as JSON.

Request errors are raised for the corresponding item unless the node is configured to continue on fail, in which case that item's output contains an error field.

Examples

Example 1: Send a prompt using incoming data

Set Messages to an OpenAI-format user message whose content references an upstream field, such as {{$json.text}}. The node sends it with the selected model and returns the API response.

Links

• n8n community nodes installation guide