ModelScope icon

ModelScope

Interact with ModelScope API-Inference services

Actions3

Large Language Model → Chat Completion

AI-generated

Summary

Sends a chat conversation to a selected ModelScope language model and returns its completion, token usage, and processing metadata. Configure one or more messages with System, User, or Assistant roles; choose a supplied message template if useful, and adjust generation settings as needed.

Inputs

  • Model (required) — Select one of the listed ModelScope language models; defaults to ZhipuAI/GLM-5.2.
  • Message Template — Choose Custom, Code Generation, Text Analysis, or Translation. The operation reads the Messages field to build the request; the source does not apply the selected template to message content.
  • Messages — Provide one or more messages, each with a System, User, or Assistant role and text content. At least one message must contain non-whitespace text.
  • Temperature — Controls output randomness from 0 to 2; defaults to 0.7.
  • Max Tokens — Maximum generated token count, from 1 to 8192; defaults to 2048.
  • Stream — Enable streaming; defaults to false.

Output shape

A single JSON item per input item.

With streaming disabled, the item contains response fields such as id, model, choices, usage, status, processing_time, input_tokens, output_tokens, total_tokens, and completed_at. With streaming enabled, it contains {"stream":true,"response":...} instead. Requires ModelScope API credentials with an access token; errors are thrown unless the node is configured to continue on fail.

Links

Discussion