chore(ct116): capture production state (LiteLLM 1.99.1, nginx /ui //docs, router decommission, trove agent)

litellm_config.yaml

@@ -47,9 +47,6 @@ litellm_settings:
   failure_callback:
   - prometheus
   model_cost:
-    gemma-4-12b:
-      input_cost_per_token: 1.5e-07
-      output_cost_per_token: 6.0e-07
     qwen3.6-27B-code:
       input_cost_per_token: 1.5e-07
       output_cost_per_token: 6.0e-07
@@ -61,7 +58,6 @@ litellm_settings:
       output_cost_per_token: 6.0e-07
   num_retries: 2
   request_timeout: 600
-  set_verbose: true
   sso_callback: /sso/callback
 model_list:
 - litellm_params:
@@ -72,16 +68,6 @@ model_list:
   model_info:
     max_input_tokens: 131072
   model_name: qwen3.6-27B-code
-- litellm_params:
-    api_base: http://192.168.68.110:8080/v1
-    api_key: not-needed
-    model: openai/gemma-4-12b
-    timeout: 120
-  model_info:
-    max_input_tokens: 131072
-    max_model_tokens: 131072
-    max_tokens: 131072
-  model_name: gemma-4-12b
 - litellm_params:
     api_base: http://192.168.68.15:8080/v1
     api_key: not-needed
@@ -102,10 +88,6 @@ model_list:
     max_input_tokens: 131072
     max_model_tokens: 131072
     max_tokens: 131072
-  model_info:
-    max_input_tokens: 131072
-    max_model_tokens: 131072
-    max_tokens: 131072
   model_name: strix-moe
 - litellm_params:
     api_base: http://192.168.68.8:8080/v1
@@ -119,9 +101,9 @@ model_list:
     max_tokens: 131072
   model_name: gpu-dense
 - litellm_params:
-    api_base: http://192.168.68.8:8080/v1
+    api_base: http://192.168.68.110:8080/v1
     api_key: not-needed
-    model: openai/qwen3.6-27B-code
+    model: openai/gpu-vision
     rpm: 500
     timeout: 300
   model_info:
@@ -137,12 +119,21 @@ model_list:
     timeout: 300
   model_info:
     max_input_tokens: 131072
-    weight: 0.55
+    max_model_tokens: 131072
+    max_tokens: 131072
+    weight: 0.70
+  model_name: syslog-auto
+- litellm_params:
+    api_base: http://192.168.68.8:8080/v1
+    api_key: not-needed
+    model: openai/qwen3.6-27B-code
+    rpm: 500
+    timeout: 300
   model_info:
     max_input_tokens: 131072
     max_model_tokens: 131072
     max_tokens: 131072
-  model_name: syslog-auto
+  model_name: qwen3.8-27B-uncensored
 - litellm_params:
     api_base: http://192.168.68.15:8080/v1
     api_key: not-needed
@@ -151,35 +142,33 @@ model_list:
     timeout: 300
   model_info:
     max_input_tokens: 131072
-    weight: 0.3
+    weight: 0.20
   model_name: syslog-auto
 - litellm_params:
     api_base: http://192.168.68.110:8080/v1
     api_key: not-needed
-    model: openai/gemma-4-12b
+    model: openai/gpu-vision
     rpm: 200
     timeout: 300
   model_info:
     max_input_tokens: 131072
-    weight: 0.15
+    weight: 0.10
   model_name: syslog-auto
 router_settings:
   allowed_fails: 100
-  enable_loadbalancing_on_proxy: false
+  enable_loadbalancing_on_proxy: true
   fallbacks:
   - syslog-auto:
     - qwen3.6-27B-code
     - strix-moe
-    - gemma-4-12b
+    - gpu-vision
   - qwen3.6-27B-code:
-    - gemma-4-12b
-  - gemma-4-12b:
-    - qwen3.6-27B-code
+    - strix-moe
   - strix-moe:
     - qwen3.6-27B-code
-    - gemma-4-12b
+    - gpu-vision
   request_timeout: 300
-  routing_strategy: usage-based-routing
+  routing_strategy: simple-shuffle
 
 agents:
 - agent_name: agent-zero-homelab