Restrict Which AI Models Your Team Can Use
When your company adopts elvex across multiple teams, you may find that not every AI model is appropriate for every user. Some models are more expensive, some are better suited for specific tasks, and some may not meet your organization's compliance or data handling requirements. Without controls, users can freely switch between any available model, making it difficult to manage costs, enforce policies, or ensure consistent quality.
elvex now gives admins the ability to block specific AI models from being used across their organization, with optional exemptions for specific roles.
You might need this feature when:
Your compliance team requires that certain models not be used with sensitive data
You want to limit access to expensive frontier models so only power users or specific teams can use them
Your organization has decided to phase out a model but wants a specific team to keep using it during a transition
You want to standardize the models your team uses to make usage analytics more consistent
How Model Restrictions Work
Admins can block specific models from the Settings > Model Access page. Once configured:
Blocked models no longer appear in the model picker for most users when starting a conversation or configuring an agent
Attempting to use a blocked model shows a clear Restricted indicator
Admins can grant specific roles an exemption, allowing users in those roles to access a blocked model even though it's restricted for everyone else
Restrictions apply both in the conversation composer and in agent configuration
Setting Up Model Restrictions
Before you begin
You must have an admin role in your elvex organization
Model restrictions are applied organization-wide
Steps
Navigate to Settings in the left sidebar to open your organization's configuration
Select Model Access from the settings menu to view the model restriction controls
Toggle on Restrict model access to enable the feature
Select the models you want to block from the model catalog — these models will no longer be available to users by default
Optionally, use the role dropdown on a blocked model to grant specific roles an exemption, allowing those roles to keep using it despite the restriction
Click Save to apply the changes — restrictions take effect immediately
What Users See
When model restrictions are active, blocked models no longer appear as selectable options in the model picker for users without an exemption. Blocked models appear with a Restricted label and cannot be selected. Users whose role has been granted an exemption for a specific model can still see and select it as normal.
Common Questions
Can I restrict models for specific teams or spaces, or only organization-wide?
Model restrictions currently apply organization-wide. All users in your organization are subject to the same set of blocked models, except those whose role has been granted an exemption for a specific model.
What happens to existing conversations or agents using a restricted model?
Existing conversations and agents that reference a blocked model will display the restriction indicator. Users without an exemption will need to switch to a permitted model to continue using that agent or start new conversations.
Can I block only a few specific models instead of managing a large list?
Yes. This feature uses a blocklist approach — you select which models to restrict, and every other model in the catalog remains available. You don't need to manage a list of every allowed model, only the ones you want to block.