UniCraft Chat Model

UniCraft AI Model Router - Universal chat completion with Smart Routing, Agent support, and multi-provider access

UniCraft Chat Model

AI-generated

Summary

Send messages to UniCraft AI chat models with options for manual model selection or automatic smart routing, supporting multiple input types and multi-modal media files, and generate various output formats including text, images, audio, video, documents, code, or markdown.

Inputs

  • Input Source (required) — Select 'Direct Input' to enter a message in the node parameters, or 'From Input Data' to use messages from incoming data items.
  • Mode (required) — Choose 'Manual' to specify a single AI model or 'Smart Routing' to let UniCraft automatically select the best model based on strategy and priority.
  • Model (required) — Pick a specific AI model to use when Mode is set to Manual.
  • Smart Routing Strategy (required) — Select the strategy for automatic model selection when Mode is 'Smart Routing'.
  • Priority (required) — Set the priority level (Low, Normal, High) for model selection in Smart Routing mode.
  • Message (required) — Enter the text message to send when using Direct Input source with text input type.
  • System Prompt — Provide an optional system prompt that customizes assistant behavior for Direct Input source.
  • Temperature — Set the sampling randomness for text generation, ranging from 0 to 2.
  • Output Type (required) — Specify the type of output to generate from the AI model, such as text, image, audio, video, document, code, or markdown.
  • Input Type (required) — Define the input type for processing, which can be text, image, audio, video, or document.
  • Media Files — Upload base64-encoded media files with MIME types and filenames for multi-modal input processing.
  • Additional Fields — Configure advanced options like maximum tokens, nucleus sampling (top_p), frequency and presence penalties, streaming response, and stop sequences.
  • Tools — Define external tools or functions the AI model can use, including their name, description, parameters schema, API endpoint, and HTTP method. Specify how tools are invoked (auto, none, required, or specific tool).

Output shape

a list of items each containing a JSON object with the AI model's reply, usage details, model identifier, cost, output and input types, and optionally media data or tool calls

The node returns an array matching input items with each item providing the assistant message content under 'response.content', tool call details if present, usage statistics, and any generated media in a 'media' field. When in Smart Routing mode, the selected model and strategy parameters are included in the output. Errors per item are returned if 'Continue On Fail' is enabled.

Examples

Example 1: Use direct input with manual model selection for a simple text chat request.

Set Input Source to 'Direct Input', Mode to 'Manual', pick a model, and provide a Message and optionally a System Prompt.

Example 2: Process multiple input data items automatically routed to the best AI model.

Set Input Source to 'From Input Data', Mode to 'Smart Routing' with chosen Strategy and Priority, and supply messages in incoming item data fields.

Example 3: Send images or other media files as input and obtain generated images or audio output.

Select Input Type and Output Type accordingly, upload media files in base64 format under Media Files, and optionally customize generation parameters.

Discussion