Update app.py

app.py

@@ -82,33 +82,38 @@ def verify_api_key(authorization: str = Header(...)):
         raise HTTPException(status_code=403, detail="Invalid or missing API key")
 
 def create_serverless_index(pc, index_name):
-    """Create a serverless index for hybrid search"""
+    """Create a Pinecone serverless index using managed model dimensions."""
     existing_indexes = [index.name for index in pc.list_indexes()]
     
     if index_name not in existing_indexes:
         pc.create_index(
             name=index_name,
-            dimension=1024,  # multilingual-e5-large dimension
-            metric="cosine"
+            dimension=1024,  # multilingual-e5-large output dimension
+            metric="cosine",
+            spec={
+                "serverless": {
+                    "cloud": "aws",   # or gcp
+                    "region": "us-west-2"
+                }
+            }
         )
         print(f"✅ Created serverless index: {index_name}")
-    else:
-        print(f"📋 Index {index_name} already exists")
     
+    # Return index object
     return pc.Index(index_name)
 
 async def embed_with_pinecone_inference(pc, texts, model="multilingual-e5-large"):
-    """Use Pinecone's hosted multilingual-e5-large model for embeddings"""
+    """Use Pinecone's hosted inference model for embeddings."""
     try:
-        # Use Pinecone Inference for embeddings
-        embeddings = pc.inference.embed(
+        inference_client = pc.inference
+        embeddings = inference_client.embed(
             model=model,
             inputs=texts,
             parameters={"input_type": "passage"}
         )
-        return [embedding['values'] for embedding in embeddings]
+        return [item["values"] for item in embeddings["results"]]
     except Exception as e:
-        print(f"❌ Embedding error: {e}")
+        print(f"❌ Pinecone Inference Error: {e}")
         raise
 
 def generate_sparse_vectors(texts):