fix(backend): rename thinking parameter to think for Ollama API compatibility

The Ollama API expects "think" parameter instead of "thinking". Also updates the API base URL in the frontend configuration to point to the correct endpoint.

backend/llm.py

@@ -81,7 +81,7 @@ async def call_ollama(prompt: str, *, tag: str = "default", temperature: float =
             },
         }
         if thinking:
-            kwargs["thinking"] = thinking
+            kwargs["think"] = thinking
 
         response = await client.chat(**kwargs)
     except Exception:
@@ -117,7 +117,7 @@ async def call_ollama(prompt: str, *, tag: str = "default", temperature: float =
     if not content.strip():
         logger.warning("[LLM][%s] empty content returned by model", tag)
 
-    return {"content": content, "thinking": thinking}
+    return {"content": content, "think": thinking}
 
 async def call_vlm_ocr(image_bytes: bytes, language: str = 'auto') -> str:
     start = time.perf_counter()

src/utils/config.js

@@ -1,6 +1,6 @@
 export const DEBUG = import.meta.env.DEV
 
-const API_BASE_URL = import.meta.env.VITE_API_BASE_URL || 'https://api.learnteach.tech:8002'
+const API_BASE_URL = import.meta.env.VITE_API_BASE_URL || 'http://149.104.29.239:8001'
 
 export const API_URL = import.meta.env.VITE_API_URL || `${API_BASE_URL}/v1/completions`
 export const OCR_URL = import.meta.env.VITE_OCR_URL || `${API_BASE_URL}/v1/ocr`