refactor(config): remove models field from vLLM endpoints (#413)

* refactor(config): remove models field from vLLM endpoints

Remove the redundant models field from VLLMEndpoint configuration.
Model-to-endpoint mapping is now solely determined by the
preferred_endpoints field in model_config, eliminating the need
for bidirectional association.

Changes:
- Remove Models field from VLLMEndpoint struct
- Update GetEndpointsForModel to use only preferred_endpoints
- Update GetAllModels to retrieve models from model_config keys
- Update all configuration files to remove models field
- Update all tests to reflect the new configuration structure
- Update TypeScript interface in dashboard frontend

This simplifies the configuration and removes potential
inconsistencies between models and preferred_endpoints.

Signed-off-by: bitliu <bitliu@tencent.com>

* more

Signed-off-by: bitliu <bitliu@tencent.com>

* more

Signed-off-by: bitliu <bitliu@tencent.com>

---------

Signed-off-by: bitliu <bitliu@tencent.com>

Author: Xunzhuo

config/config-mcp-classifier-example.yaml

@@ -75,8 +75,6 @@ vllm_endpoints:
   - name: endpoint1
     address: 127.0.0.1
     port: 8000
-    models:
-      - openai/gpt-oss-20b
     weight: 1
     health_check_path: /health
 

config/config.development.yaml

@@ -29,8 +29,6 @@ vllm_endpoints:
   - name: "local-endpoint"
     address: "127.0.0.1"
     port: 8000
-    models:
-      - "test-model"
     weight: 1
 
 model_config:

config/config.e2e.yaml

@@ -42,15 +42,11 @@ vllm_endpoints:
   - name: "qwen-endpoint"
     address: "127.0.0.1"
     port: 8000
-    models:
-      - "Model-A"
     weight: 1
     health_check_path: "/health"
   - name: "tinyllama-endpoint"
     address: "127.0.0.1"
     port: 8001
-    models:
-      - "Model-B"
     weight: 1
     health_check_path: "/health"
 

config/config.production.yaml

@@ -34,8 +34,6 @@ vllm_endpoints:
   - name: "endpoint1"
     address: "127.0.0.1"
     port: 8000
-    models:
-      - "openai/gpt-oss-20b"
     weight: 1
 
 model_config:

config/config.recipe-accuracy.yaml

@@ -44,8 +44,6 @@ vllm_endpoints:
   - name: "endpoint1"
     address: "127.0.0.1"
     port: 8000
-    models:
-      - "openai/gpt-oss-20b"
     weight: 1
 
 model_config:

config/config.recipe-latency.yaml

@@ -39,8 +39,6 @@ vllm_endpoints:
   - name: "endpoint1"
     address: "127.0.0.1"
     port: 8000
-    models:
-      - "openai/gpt-oss-20b"
     weight: 1
 
 model_config:

config/config.recipe-token-efficiency.yaml

@@ -44,8 +44,6 @@ vllm_endpoints:
   - name: "endpoint1"
     address: "127.0.0.1"
     port: 8000
-    models:
-      - "openai/gpt-oss-20b"
     weight: 1
 
 model_config:

config/config.testing.yaml

@@ -30,8 +30,6 @@ vllm_endpoints:
   - name: "mock"
     address: "172.28.0.10"
     port: 8000
-    models:
-      - "openai/gpt-oss-20b"
     weight: 1
     health_check_path: "/health"
 

config/config.yaml

@@ -34,8 +34,6 @@ vllm_endpoints:
   - name: "endpoint1"
     address: "127.0.0.1"  # IPv4 address - REQUIRED format
     port: 8000
-    models:
-      - "openai/gpt-oss-20b"
     weight: 1
 
 model_config:

config/examples/system_prompt_example.yaml

@@ -100,9 +100,9 @@ default_reasoning_effort: medium
 # vLLM endpoints configuration
 vllm_endpoints:
   - name: "mock"
-    address: "http://127.0.0.1:8000"
-    models:
-      - "openai/gpt-oss-20b"
+    address: "127.0.0.1"
+    port: 8000
+    weight: 1
 
 # Usage Notes:
 # 1. System prompts are automatically injected based on query classification