Models & routing

Roseram supports multi-model routing: you can use Roseram’s routing and/or select configured models, and adjust reasoning effort where the product exposes that control.

Goals

  • Keep model choice next to the plan → build → review loop
  • Reduce hopping across unrelated chat tabs
  • Make cost visible when you pick heavier models (see the usage dashboard)

How to think about routing

  1. Default / Roseram routing — let the product route for the task class when that option is available.
  2. Manual selection — pick a configured model when you know you need a specific provider behavior.
  3. Reasoning effort — increase for hard design/debug tasks; decrease for routine edits when quality allows.

Exact model names and availability change with configuration and capacity. This doc does not freeze a permanent model list — check the in-app model picker.

Availability caveats (honest)

  • Not every model is always on
  • Providers rate-limit; Roseram may surface capacity or safety limits
  • A “better” model is not free — watch the usage meter

Tips

  • Start with a mid-tier route for exploration; escalate effort only when stuck
  • For reviews, prefer clarity over maximum spend
  • If a model fails or is unavailable, switch routes and retry — then troubleshoot if it persists

Related

Try routing on a real task: https://roseram.com/launch

---

Ready to try Roseram?

Start free trial →