17 views
-/https://github.com/berriai/litellm/issues/24734
GitHub · issue

#24734 [Bug]: False "guardrails feature unavailable" error when guardrails are not configured in LiteLLM 1.82.3

  • State: open
  • Author: @ianustec
  • Labels: bug, ui-dashboard

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

After upgrading LiteLLM to version 1.82.3, editing any API key configuration (even for models without guardrails) triggers a false "Feature Unavailable" error claiming the guardrails feature requires a LiteLLM Enterprise license. The error persists despite having no guardrails configured in the application or environment.

<img width="1773" height="940" alt="Image" src="https://github.com/user-attachments/assets/39dbc0fa-4c55-4536-a077-8c329b888a23" />

`

### Steps to Reproduce

1. Install LiteLLM 1.82.3. 2. Configure a model (e.g., litellm.completion(model="gpt-4", api_key="xxx")) without guardrails. 3. Attempt to edit the API key via UI or configuration method. 4. Observe the error: `{'error': 'This feature is only available for LiteLLM Enterprise users: guardrails...'}`

### Relevant log output

```shell

```

### What part of LiteLLM is this about?

UI Dashboard

### What LiteLLM version are you on ?

v1.82.3

### Twitter / LinkedIn details

_No response_

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (2 events)
#0 of 0 · 31d19h4m59s ago — entered · #import:https:::github.com:berriai:litellm post #2105
The right issue is harder because it requires provider-specific request translation, precedence handling between related parameters, compatibility across model variants, and regression coverage. The left issue appears more localized to an incorrect dashboard entitlement check and should be lower-risk to isolate and correct.
#0 of 0 · 31d19h15s ago — current · #import:https:::github.com:berriai:litellm post #2183
Provider-specific response translation and validation across both streaming and non-streaming tool-call paths creates broader compatibility and regression risk; the dashboard issue is more likely a localized entitlement-check regression.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search