Model Selector

Pick one of several connected chat models at runtime based on workflow data. Max 10 models; OpenAI is not available on the platform.

Written By Philippe

Last updated 2 days ago

Model Selector

✅ Supported — with the limitations below

What It Does

Chooses one of several connected chat models during a run, based on rules you define. Use it for fallback (try a cheaper model first, escalate on failure) or task routing (pick a model per input).

Connections

InputRequiredDescription
Model 1…NAt least oneThe chat models to choose between

Output: a single Chat Model connection — wire it into an Agent or chain exactly like a normal chat model.

How It Works

  1. Rules are evaluated top to bottom.
  2. The first rule whose conditions match selects its model.
  3. That model is handed to the parent node for the run.
  4. If no rule matches, the run fails with No matching rule found.

Compatibility

FeatureStatus
Rule-based model selectionSupported
Fallback / task routingSupported
Number of modelsUp to 10 — the maximum the node allows
Routing to OpenAI modelsNot available — OpenAI chat models are not supported on the platform. Route between Anthropic, Google Gemini, OpenRouter, Ollama, DeepSeek, xAI Grok, Groq, or Mistral instead

Notes

  • The builder always shows 10 model inputs; leave the unused ones unconnected.
  • Selection is based on your rules only — it does not measure model latency or cost automatically.