Routing Presets — ML Junction docs

Preset · What it changes · Recommended use
stable_production · Quality strategy; preserves context and privacy · Production agents and repeatable workflows
lowest_cost · Cost strategy; same-or-lower-price fallback · Batch extraction and cost-sensitive workloads
lowest_latency · Latency strategy · Interactive chat and latency-sensitive APIs
strongest_privacy · Requires ZDR and privacy-preserving fallback · Sensitive workloads
long_context · Preserves context-window capacity · Large documents and long conversations

{
  "model": "gpt-5.5",
  "messages": [
    { "role": "user", "content": "Review this confidential contract." }
  ],
  "routing": {
    "preset": "strongest_privacy",
    "allow_provider_training": false
  }
}

A preset supplies coherent defaults but does not replace explicit requirements. For example, strongest_privacy controls route privacy; it does not automatically require PDF support for a file upload.

Canonical URL: https://mljunction.com/docs/routing-presets