Skip to main content

Allowed Models

V2 console

This card lives in Policy Engine > LLM Gateway at web.quilr.ai/policy. Edits join the shared draft and take effect once you review and publish a revision.

Choose which models matching traffic may call, and which models nobody in scope may call. A request for a model outside the effective list is rejected at the gateway.

Allowed Models card collapsed, showing the Allowed models and Rejected models summary rows

How lists combine​

Allowed Models card expanded with the combine banner, the Allowed models section and its Add allowed list button

  • Allowed lists combine. Every allowed list that matches a request is merged, and the request may use any model in the union.
  • Rejected models always win, including models reached through routing. Priority is ignored for both lists.
  • No matching allowed list means any configured model may be requested unless it is rejected.
  • Other policies still apply. Allowing a model does not bypass identity, access, budget or limit checks.

Sections​

SectionWhat it doesEmpty state
Allowed modelsRestricts matching requests to the listed models. Add allowed list.Any configured model may be requested.
Rejected modelsBlocks the listed models for matching requests, even when another list allows them. Add rejected list.No model is explicitly rejected.

Rejected models section with the Add rejected list button and the empty state

Both sections apply on assistants, bedrock, chat, copilot, embeddings, models, realtime, rerank, responses, sdk_check, stt, text, tts and vertex.

Configuration settings​

Both buttons open the same New Allowed Models configuration dialog. Leave Allowed models empty to add only rejections.

Model picker grouped by provider credential, with search, a selected count, Select all and per-group All and None

SettingOptionsDefaultNotes
Applies toEveryone, People, Smart group, Application, App tag, Provider, API surface, Environment, Prompt complexity, Prompt text, Tool, Source network, Except...EveryoneOr pick an application directly. A pencil on an app means it already has a configuration; picking it opens that one.
Allowed modelsCheckboxes grouped by provider credential, with Search models, Select all, Clear all and per-group All / NoneNoneModels come from the provider credentials configured under Settings > LLM Gateway. Only enabled credentials are listed.
Rejected modelsSame pickerNone
SeverityNot set, Very low, Low, Medium, High, Critical, Very criticalNot setReported on matching requests for dashboards, exports and alerts. Never changes the outcome.
More optionsOpen in full editor-Priority, extra rules, metadata and content conditions, raw QuilrQL.

Reads as, Severity and More options rows with the Add configuration button

Everyone lists affect every app

An allowed list scoped to Everyone restricts every application and user to that list. Choose Application or another scope to limit who it affects.

Examples​

support_copilot_modelsrequest

runs on requestpriority 500

WhenApplicationisSupport Copilot
Then
Allowed modelsgpt-4.1gpt-4.1-mini

Support Copilot may only call the two listed models. A request for gpt-4o is rejected.

reject_preview_modelsrequest

runs on requestpriority 500

Whenalways
Then
Rejected modelso1-previewgpt-4.5-preview

No one may call the preview models, even through an app whose allowed list includes them or a routing group that targets them.

Scoping and precedence​

  • New configurations start at priority 500. Priority does not matter on this card: allowed lists union and rejections always win.
  • The card runs before routing. Allowed Models filters what may be used, then Routing Groups & Fallbacks picks where the request goes.

Legacy app setting​

An app's usable models come from the providers linked to it. See Link providers to an app.

On this page