fix: lmstudio tests and integration

Co-authored-by: Jin Peng <jinjpeng@Jins-MacBook-Pro.local>

letta/schemas/providers/lmstudio.py

@@ -55,7 +55,7 @@ class LMStudioOpenAIProvider(OpenAIProvider):
                 LLMConfig(
                     model=model_name,
                     model_endpoint_type="openai",
-                    model_endpoint=self.base_url,
+                    model_endpoint=self.model_endpoint_url,
                     context_window=context_window_size,
                     handle=self.get_handle(model_name),
                     compatibility_type=compatibility_type,
@@ -94,7 +94,7 @@ class LMStudioOpenAIProvider(OpenAIProvider):
                 EmbeddingConfig(
                     embedding_model=model_name,
                     embedding_endpoint_type="openai",
-                    embedding_endpoint=self.base_url,
+                    embedding_endpoint=self.model_endpoint_url,
                     embedding_dim=768,  # Default embedding dimension, not context window
                     embedding_chunk_size=DEFAULT_EMBEDDING_CHUNK_SIZE,  # NOTE: max is 2048
                     handle=self.get_handle(model_name),

tests/configs/llm_model_configs/lmstudio.json

@@ -2,7 +2,7 @@
   "context_window": 8192,
   "model": "qwen2.5-7b-instruct-1m",
   "model_endpoint_type": "openai",
-  "model_endpoint": "http://127.0.0.1:1234/v1",
+  "model_endpoint": "http://127.0.0.1:1234/api/v0",
   "model_wrapper": null,
   "provider_name": "lmstudio_openai"
 }

tests/integration_test_send_message.py

@@ -110,11 +110,7 @@ USER_MESSAGE_BASE64_IMAGE: List[MessageCreate] = [
 ]
 
 # configs for models that are to dumb to do much other than messaging
-limited_configs = [
-    "ollama.json",
-    "together-qwen-2.5-72b-instruct.json",
-    "vllm.json",
-]
+limited_configs = ["ollama.json", "together-qwen-2.5-72b-instruct.json", "vllm.json", "lmstudio.json"]
 
 all_configs = [
     "openai-gpt-4o-mini.json",