Fix: prevent double embedding in mem0.add (fixes #3723)

mem0/memory/main.py

@@ -407,7 +407,8 @@ def _add_to_vector_store(self, messages, metadata, filters, infer):
 
                 msg_content = message_dict["content"]
                 msg_embeddings = self.embedding_model.embed(msg_content, "add")
-                mem_id = self._create_memory(msg_content, msg_embeddings, per_msg_meta)
+                # Pass embeddings as a mapping so _create_memory reuses the computed vector
+                mem_id = self._create_memory(msg_content, {msg_content: msg_embeddings}, per_msg_meta)
 
                 returned_memories.append(
                     {
@@ -1074,8 +1075,11 @@ def history(self, memory_id):
 
     def _create_memory(self, data, existing_embeddings, metadata=None):
         logger.debug(f"Creating memory with {data=}")
-        if data in existing_embeddings:
+        # existing_embeddings may be a mapping (preferred) or a precomputed vector
+        if isinstance(existing_embeddings, dict) and data in existing_embeddings:
             embeddings = existing_embeddings[data]
+        elif existing_embeddings is not None and not isinstance(existing_embeddings, dict):
+            embeddings = existing_embeddings
         else:
             embeddings = self.embedding_model.embed(data, memory_action="add")
         memory_id = str(uuid.uuid4())
@@ -1437,7 +1441,8 @@ async def _add_to_vector_store(
 
                 msg_content = message_dict["content"]
                 msg_embeddings = await asyncio.to_thread(self.embedding_model.embed, msg_content, "add")
-                mem_id = await self._create_memory(msg_content, msg_embeddings, per_msg_meta)
+                # Pass embeddings as a mapping so _create_memory reuses the computed vector
+                mem_id = await self._create_memory(msg_content, {msg_content: msg_embeddings}, per_msg_meta)
 
                 returned_memories.append(
                     {
@@ -2135,8 +2140,11 @@ async def history(self, memory_id):
 
     async def _create_memory(self, data, existing_embeddings, metadata=None):
         logger.debug(f"Creating memory with {data=}")
-        if data in existing_embeddings:
+        # existing_embeddings may be a mapping (preferred) or a precomputed vector
+        if isinstance(existing_embeddings, dict) and data in existing_embeddings:
             embeddings = existing_embeddings[data]
+        elif existing_embeddings is not None and not isinstance(existing_embeddings, dict):
+            embeddings = existing_embeddings
         else:
             embeddings = await asyncio.to_thread(self.embedding_model.embed, data, memory_action="add")
 

mem0/vector_stores/azure_ai_search.py

@@ -222,7 +222,8 @@ def search(self, query, vectors, limit=5, filters=None):
         if filters:
             filter_expression = self._build_filter_expression(filters)
 
-        vector_query = VectorizedQuery(vector=vectors, k_nearest_neighbors=limit, fields="vector")
+        # Azure VectorizedQuery uses `k` for nearest neighbors in current SDKs.
+        vector_query = VectorizedQuery(vector=vectors, k=limit, fields="vector")
         if self.hybrid_search:
             search_results = self.search_client.search(
                 search_text=query,

tests/test_memory.py

@@ -245,3 +245,35 @@ def test_get_all_handles_flat_list_from_postgres(mock_sqlite, mock_llm_factory,
     assert len(result) == 2
     assert result[0]["memory"] == "Memory 1"
     assert result[1]["memory"] == "Memory 2" 
+
+
+@patch('mem0.utils.factory.EmbedderFactory.create')
+@patch('mem0.utils.factory.VectorStoreFactory.create')
+@patch('mem0.utils.factory.LlmFactory.create')
+@patch('mem0.memory.storage.SQLiteManager')
+def test_add_infer_false_embeds_once(mock_sqlite, mock_llm_factory, mock_vector_factory, mock_embedder_factory):
+    """
+    Regression test: adding with infer=False should not trigger duplicate embedding calls.
+    """
+    embedder = MagicMock()
+    embedder.embed.return_value = [0.1, 0.2, 0.3]
+    embedder.config = MagicMock(embedding_dims=3)
+    mock_embedder_factory.return_value = embedder
+
+    mock_vector_store = MagicMock()
+    mock_vector_store.search.return_value = []
+    mock_vector_store.insert.return_value = None
+    mock_vector_store.get.return_value = None
+    telemetry_vector_store = MagicMock()
+    mock_vector_factory.side_effect = [mock_vector_store, telemetry_vector_store]
+
+    mock_llm_factory.return_value = MagicMock()
+    mock_sqlite.return_value = MagicMock()
+
+    from mem0.memory.main import Memory as MemoryClass
+    memory = MemoryClass(MemoryConfig())
+
+    memory.add("foo", user_id="test_user", infer=False)
+
+    assert embedder.embed.call_count == 1
+    mock_vector_store.insert.assert_called_once()

tests/vector_stores/test_azure_ai_search.py

@@ -542,7 +542,7 @@ def test_search_basic(azure_ai_search_instance):
     # Check parameters
     assert len(kwargs["vector_queries"]) == 1
     assert kwargs["vector_queries"][0].vector == query_vector
-    assert kwargs["vector_queries"][0].k_nearest_neighbors == 5
+    assert kwargs["vector_queries"][0].k == 5
     assert kwargs["vector_queries"][0].fields == "vector"
     assert kwargs["filter"] is None  # No filters
     assert kwargs["top"] == 5