Panda Free LLM icon

Panda Free LLM

Send a prompt to a free LLM provider, with automatic failover to other free providers when one is rate-limited or out of quota

Panda Free LLM

AI-generated

Summary

Send each input item’s system and user prompts to the selected free LLM provider, trying configured fallback provider/model pairs in order when needed. Return the reply, provider and model used, token usage when supplied, and the attempt history; JSON mode also parses the reply when possible.

Inputs

  • System Prompt — Fixed instructions sent as the system message. Defaults to “You are a helpful assistant.”
  • User Prompt (required) — Prompt sent for each input item; supports expressions such as mapping a field from the previous node.
  • Response Format — Choose Text or JSON. JSON mode asks for a JSON object and attempts to parse the reply into outputParsed.
  • Provider — Primary provider to try: Groq, Cerebras, Gemini, OpenRouter, or Mistral. Defaults to Groq.
  • Model — Model for the primary provider. If empty, uses that provider’s default model.
  • Enable Failover — When enabled, tries the configured fallback models in order after the primary attempt. Defaults to enabled.
  • Fallback Models (in order) — Ordered provider/model choices attempted after the primary; rows without a selection and duplicate choices are skipped.
  • Options — Set maximum tokens (default 1024), temperature (default 0.7), per-provider timeout in milliseconds (default 30000), and whether to throw if every provider fails (default true).

Output shape

One output item per input item that receives a reply, or per item when all-provider failures are configured to continue.

Successful items contain output (reply text), provider, model, usage (or null), and attempts. In JSON mode, outputParsed contains the parsed value or null with parseError if parsing fails. If all attempts fail and Throw If All Providers Fail is disabled, the item contains output: null, success: false, and attempts. Otherwise the node errors; n8n’s Continue On Fail setting can instead produce an item with an error field. Providers without configured API keys are recorded as skipped attempts.

Links

Discussion