use my local url and api-key

[dacai-academic]

I'm using a local vLLM instance with multiple models, and /v1/models is working correctly. The problem is that when I set the base_url and api_key in the UI, it doesn't auto-populate the model list. It looks like it might have trouble with local URLs. Here’s my setup:
librechat.yaml

version: 1.3.5
cache: true
endpoints:
  custom:
    - name: "vLLM"
      apiKey: "user_provided"
      baseURL: "user_provided"
      models:
        default: [""]
        fetch: true
      titleConvo: true
      titleModel: "current_model"
      dropParams: ["stop"]
      modelDisplayLabel: "vLLM"

**.env**

env.txt

[dacai-academic]

my test url: http:192.168.100.188:8089/v1
model_list: curl http://192.168.100.188:8089/v1/models
{"object":"list","data":[{"created":1776417619,"id":"glm-5.1","object":"model","owned_by":"system"},{"created":1776417619,"id":"glm-5.0","object":"model","owned_by":"system"},{"created":1776417619,"id":"claude-opus-4-7","object":"model","owned_by":"system"},{"created":1776417619,"id":"claude-opus-4-6","object":"model","owned_by":"system"},{"created":1776417619,"id":"gemini-3.1-pro-preview","object":"model","owned_by":"system"},{"created":1776417619,"id":"gpt-5.4","object":"model","owned_by":"system"},{"created":1776417619,"id":"gemini-3.1-flash-image-preview","object":"model","owned_by":"system"},{"created":1776417619,"id":"gemini-3-pro-image-preview","object":"model","owned_by":"system"}]}

[francoisauclair911]

Having similar issues when using helm chart on EKS (which reference the url as local url. the fetch: true doesn't seem to apply.
However it only bugs when apiKey: "user_provided", when we use env variables it works.

local url looks like
"http://litellm.infra-tools.svc.cluster.local:4000/v1"