LiteLLM: fix model routing config

apps/litellm/README.md

@@ -1,14 +1,19 @@
 # LiteLLM
 
 LiteLLM is the cluster-local LLM proxy for OpenWebUI. It routes requests to the
-external llama.cpp OpenAI-compatible endpoint at `http://framework-llm:8080/v1`.
+external llama.cpp OpenAI-compatible endpoint at
+`http://framework-llm.open-webui.svc.cluster.local:8080/v1`.
 
 The public dashboard is exposed at `https://litellm.noxxos.nl` through
 LiteLLM's native Authentik SSO. API clients should use LiteLLM bearer keys.
 
 ## Secrets
 
-Create the LiteLLM secret before syncing:
+Create the LiteLLM secrets before syncing. The Helm chart references these
+secrets directly, so the LiteLLM pod will fail with `secret "litellm-sso" not
+found` until the SSO secret exists in the `litellm` namespace.
+
+Create the LiteLLM master key and salt:
 
 ```bash
 kubectl create namespace litellm
@@ -58,7 +63,7 @@ if ak_is_group_member(request.user, name="admins"):
 return {"litellm_role": "internal_user"}
 ```
 
-Create the LiteLLM SSO secret:
+Then create the LiteLLM SSO secret before syncing the app:
 
 ```bash
 kubectl -n litellm create secret generic litellm-sso \

apps/litellm/values.yaml

@@ -51,27 +51,40 @@ litellm:
 
   proxy_config:
     model_list:
-      - model_name: qwen3-27b
+      - model_name: qwen27
         litellm_params:
-          model: openai/qwen3.6-27B
-          api_base: http://framework-llm:8080/v1
+          model: openai/qwen27
+          api_base: http://framework-llm.open-webui.svc.cluster.local:8080/v1
           api_key: not-needed
         model_info:
           mode: chat
-      - model_name: qwen3-122b-a10b
+          max_input_tokens: 131072
+      - model_name: qwen122
         litellm_params:
-          model: openai/qwen3.5-122B-A10B
-          api_base: http://framework-llm:8080/v1
+          model: openai/qwen122
+          api_base: http://framework-llm.open-webui.svc.cluster.local:8080/v1
           api_key: not-needed
         model_info:
           mode: chat
+          max_input_tokens: 262144
+          max_output_tokens: 16384
+      - model_name: qwen122-abliterated
+        litellm_params:
+          model: openai/qwen122-abliterated
+          api_base: http://framework-llm.open-webui.svc.cluster.local:8080/v1
+          api_key: not-needed
+        model_info:
+          mode: chat
+          max_input_tokens: 262144
+          max_output_tokens: 16384
       - model_name: qwen35-4b
         litellm_params:
           model: openai/qwen35-4b
-          api_base: http://framework-llm:8080/v1
+          api_base: http://framework-llm.open-webui.svc.cluster.local:8080/v1
           api_key: not-needed
         model_info:
           mode: chat
+          max_input_tokens: 32768
     general_settings:
       master_key: os.environ/PROXY_MASTER_KEY
     litellm_settings:

apps/open-webui/README.md

@@ -4,7 +4,7 @@ OpenWebUI is exposed at `https://openwebui.noxxos.nl` and connects to LiteLLM at
 `http://litellm.litellm.svc.cluster.local:4000/v1`.
 
 LiteLLM then routes requests to the OpenAI-compatible llama.cpp endpoint at
-`http://framework-llm:8080/v1`.
+`http://framework-llm.open-webui.svc.cluster.local:8080/v1`.
 
 ## Authentik