Visitar URL original
feat: Allow LiteLLM backends to utilize filter functions · Issue #33337 · open-webui/open-webui · GitHub
Skip to content

feat: Allow LiteLLM backends to utilize filter functions #33337

Open

Description

@treydock

Before Submitting

  • I searched open and closed issues and discussions for an existing request.
  • I checked whether this already exists on the dev branch or latest source.
  • I understand that maintainers want a well-written issue or discussion before any code pull request.
  • This request is not a security vulnerability.

Problem

This is similar to #32672 but for LiteLLM connections. Right now we have one LiteLLM connection that is our parent org , where we get a single Bearer token for our whole org to access models like Opus 5. We rely on filter functions to handle our accounting through OTEL metrics but we've noticed if I set the Connection in Open WebUI to be Provider=LiteLLM then none of our inlet or outlet functions are called.

Desired Behavior

I'd like Anthropic passthrough with LiteLLM to still call filter functions since we need our organization's accounting and metrics to identify who in our org is using Opus 5 while LiteLLM just shows our shared hpcgpt token as the usage so our Open WebUI metrics give us details into who's using the shared resource.

Why This Matters

We don't manage the LiteLLM we're using and we have a single shared bearer token for our entire org's usage so even if we rely on LiteLLM reporting, it's all just the same user and there are limits as we don't manage the parent organization's LiteLLM. We rely on our organization's metrics to identify who in our org is using Opus 5 through the Open WebUI connection.

Examples or References

No response

Alternatives or Workarounds

For now we've set Provider=Default but seems like Provider=LiteLLM on the connection should still allow us to use filter functions.

Activity

  1. owui-terminator commented on Oct 28, 2026

    @owui-terminator

    🔍 Related Issues Found

    I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions:

    1. 🟣 #32672 feat: Allow OpenAI embeddings to utilize filter functions
      This is the closest analogue: it requests enabling filter functions for a specific backend (OpenAI embeddings), and the same accounting/metrics motivation is described. The new issue is the LiteLLM version of that request.
      by treydock · enhancement

    2. 🟣 #3237 Enh: call outlet hook from the backend
      This issue reports that when Open WebUI is used as an API proxy, only inlet hooks run and outlet hooks are skipped. The new LiteLLM request is about backend/provider-specific behavior preventing filter hooks from running, which is the same underlying class of problem.
      by frederikschubert

    3. 🟣 #26244 bug: custom Filter inlet() is called but outlet() is not called after GitHub update in install
      This confirms a broader filter-hook bug where inlet is called but outlet is not, which matches the user’s complaint that LiteLLM connections do not trigger filter functions. It is not LiteLLM-specific, but it is directly related to the hook execution issue.
      by etiennegnome · bug


    💡 If your issue is a duplicate, please close it and add any additional details to the existing issue instead.

    This comment was generated automatically. React with 👍 if helpful, 👎 if not.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions