Model Access

Choose which model tiers everyone in your organization may call, and whether tiers added later arrive switched on or off

People in your organization don't call a provider's model directly. They call a model tier by name, such as the one a coding agent is configured to use, and Dispatch sends the request to a model behind it. The Models tab on the Organization page decides which of those tiers anyone in your organization may call. It applies to every key in the organization at once.

Two pages called Models: This tab is where a Dispatch Admin decides what's allowed. The Models page in the main navigation is read-only, and it's where everyone, including Dispatch Users, goes to see what they may call. A tier you switch off here stays listed there, marked as "disabled by your organization", so people find an answer instead of a missing tier.

Why It's Organization-Wide

Model access is set for the whole organization, not per person. A tier name is what goes into a shared configuration file for a coding agent, so a team's shared setup works the same for everybody. To keep a model from a particular group, use a guardrail, which can switch off tiers for the people under it but never switch on a tier the organization has switched off.

Which Tiers You See

The card, Models your team can call, lists every tier DarcyIQ currently offers your organization, each with its display name and the exact name tools use to request it. If no tiers are on offer yet, the card says "No models are available to this organization yet."

Your data residency setting can narrow this list. A tier with no models left in your chosen region isn't offered.

Switching Tiers On and Off

1

Use the switch beside each tier A tier that's switched on can be called by every key in the organization. One that's switched off is refused if anyone asks for it by name.

2

Choose how new tiers arrive Under Models Darcy adds later, choose Available to my team or Off until I switch them on. This decides what a tier published after today arrives as. The choices you've already made above are kept either way.

3

Click Save models The count beside the button shows how many tiers are available to your team, for example "3 of 5 available to your team." The change applies to every key in the organization within a minute.

Leave at least one model on. With every tier switched off, every request from the organization would be refused, so Dispatch won't let you save that.

The Auto Tier

A tier marked picks for you is Auto. It "Keeps choosing between models for each request, including ones switched off here — switching one off stops your team asking for it by name." This is what makes running on Auto alone work: you can switch off every other tier, and Auto still has models to choose from.

What People See When a Tier Is Off

• In their tools: A switched-off tier is left out of the model list that tools fetch from Dispatch, so it doesn't appear in a model picker.

• When asked for by name: The request is refused with a message that the tier "is disabled for your organization", asking them to contact their administrator and listing the models they can use instead.

• On the Models page: The tier stays listed, marked as disabled by the organization.

Next Steps

Goal	Documentation
See what your team can call	Models
Switch off models for one group of people	Guardrails
Restrict where requests are processed	Data Residency
Connect a tool to a tier	Cookbooks