LLM Assignment

Purpose

Assign specific LLMs to individual users or groups so mentor editors only see the models they are authorized to use, rather than every model available to tenant admins.

---

Prerequisites

• The user must already have mentor editor access (visible on the mentor's Access tab)
• A role with LLM-related actions must exist (or you will create one)

---

Create an LLM Access Role

1. Go to Tenant Settings → Management → Roles.
2. Create a new role (e.g., "LLM Model Access").
3. Assign the actions:
   • Read all LLMs
   • Select from available LLMs

---

Create or Edit a Policy

1. Navigate to the Policies tab.
2. Click New Policy (or edit an existing one like "LLM Model Access").
3. Configure the policy:
   • Policy name: descriptive label
   • Role: select the LLM access role created above
   • Platform: select your tenant
   • LLM(s): choose one or more models (e.g., OpenAI GPT-4o Mini, Google, Perplexity)
4. Under Users/Groups, add the specific user or group.
5. Click Save Policy.

---

Verify as the User

1. Log in as the assigned user.
2. Open a mentor and click the LLMs tab.
3. Confirm that only the assigned models appear (e.g., GPT-4o Mini, Google, Perplexity).
4. Tenant admins will still see all LLMs; restricted users see only their assigned models.

---

Scope Options

• Platform-wide: the user sees the same limited LLM set across all mentors
• Per-mentor: restrict LLM access to specific mentors by adjusting the policy resource to a particular mentor rather than the whole platform

---

Key Takeaways

• Roles define what actions can be taken (read, select LLMs)
• Policies bind a role to specific LLM resources and assign users or groups
• LLM restrictions apply immediately upon the user's next login
• Use groups to manage LLM access for multiple users at once