Skip to main content
A router can only select from its enabled model list. Keep that list focused: include the models you are willing to pay for and trust for the workload behind that router.

​
Model ID Format

Dari Router uses provider-prefixed model IDs:
ProviderPrefixExample
OpenAIopenai/openai/gpt-5.5
Anthropicanthropic/anthropic/claude-sonnet-4-6
OpenRouteropenrouter/openrouter/google/gemini-3.1-pro-preview
The provider prefix determines which provider key source Dari uses after the router selects a model. That key source can be Dari-managed or a BYOK key stored on the router.

​
Example Model Set

[
  "openai/gpt-5.5",
  "openai/o4-mini",
  "anthropic/claude-sonnet-4-6",
  "anthropic/claude-haiku-4-5",
  "openrouter/google/gemini-3.1-pro-preview"
]
This set gives the router a mix of stronger and faster models across providers.

​
Update Models

Update the model list from the router detail page in the Dari dashboard. Removing a model takes it out of the router’s candidate set for future requests. Adding a model may require choosing a provider key source if the model uses a provider that is not already configured on the router.

​
Best Practices

Use separate routers for materially different workloads. A customer-facing chat router, an internal batch-processing router, and a coding-assistant router usually deserve different model sets and provider key sources.