From 8f3b60f3e1a973ecea667a3019eebedd96cfe688 Mon Sep 17 00:00:00 2001
From: Hrushikesh Yadav <136978914+HrushiYadav@users.noreply.github.com>
Date: Fri, 12 Jun 2026 20:41:18 +0530
Subject: [PATCH] fix: prevent crash in parse_vision_messages when vision is
 disabled (#5487)

---
 mem0/memory/utils.py              | 17 ++++++---
 tests/memory/test_memory_utils.py | 57 ++++++++++++++++++++++++++++++-
 2 files changed, 69 insertions(+), 5 deletions(-)

diff --git a/mem0/memory/utils.py b/mem0/memory/utils.py
index f03341efe..253f6b4b4 100644
--- a/mem0/memory/utils.py
+++ b/mem0/memory/utils.py
@@ -179,11 +179,20 @@ def parse_vision_messages(messages, llm=None, vision_details="auto"):
 
         # Handle message content
         if isinstance(msg["content"], list):
-            # Multiple image URLs in content
-            description = get_image_description(msg, llm, vision_details)
-            returned_messages.append({"role": msg["role"], "content": description})
+            if llm is None:
+                text_parts = [
+                    part["text"] for part in msg["content"]
+                    if isinstance(part, dict) and part.get("type") == "text"
+                ]
+                if not text_parts:
+                    continue
+                returned_messages.append({"role": msg["role"], "content": " ".join(text_parts)})
+            else:
+                description = get_image_description(msg, llm, vision_details)
+                returned_messages.append({"role": msg["role"], "content": description})
         elif isinstance(msg["content"], dict) and msg["content"].get("type") == "image_url":
-            # Single image content
+            if llm is None:
+                continue
             image_url = msg["content"]["image_url"]["url"]
             try:
                 description = get_image_description(image_url, llm, vision_details)
diff --git a/tests/memory/test_memory_utils.py b/tests/memory/test_memory_utils.py
index 41a02f76d..8cf39475a 100644
--- a/tests/memory/test_memory_utils.py
+++ b/tests/memory/test_memory_utils.py
@@ -1,5 +1,60 @@
 import pytest
-from mem0.memory.utils import remove_spaces_from_entities, sanitize_relationship_for_cypher
+from unittest.mock import Mock
+
+from mem0.memory.utils import parse_vision_messages, remove_spaces_from_entities, sanitize_relationship_for_cypher
+
+
+class TestParseVisionMessages:
+    def test_multimodal_list_without_llm_extracts_text(self):
+        messages = [
+            {"role": "user", "content": [
+                {"type": "text", "text": "What is this?"},
+                {"type": "image_url", "image_url": {"url": "https://example.com/img.png"}},
+            ]},
+        ]
+        result = parse_vision_messages(messages, llm=None)
+        assert len(result) == 1
+        assert result[0]["role"] == "user"
+        assert result[0]["content"] == "What is this?"
+
+    def test_image_dict_without_llm_is_skipped(self):
+        messages = [
+            {"role": "user", "content": {"type": "image_url", "image_url": {"url": "https://example.com/img.png"}}},
+            {"role": "user", "content": "hello"},
+        ]
+        result = parse_vision_messages(messages, llm=None)
+        assert len(result) == 1
+        assert result[0]["content"] == "hello"
+
+    def test_multimodal_with_llm_calls_generate_response(self):
+        mock_llm = Mock()
+        mock_llm.generate_response.return_value = "A photo of a cat"
+        messages = [
+            {"role": "user", "content": [
+                {"type": "text", "text": "Describe this"},
+                {"type": "image_url", "image_url": {"url": "https://example.com/cat.png"}},
+            ]},
+        ]
+        result = parse_vision_messages(messages, llm=mock_llm, vision_details="auto")
+        assert result[0]["content"] == "A photo of a cat"
+        mock_llm.generate_response.assert_called_once()
+
+    def test_image_only_list_without_llm_is_skipped(self):
+        messages = [
+            {"role": "user", "content": [
+                {"type": "image_url", "image_url": {"url": "https://example.com/img.png"}},
+            ]},
+        ]
+        result = parse_vision_messages(messages, llm=None)
+        assert result == []
+
+    def test_plain_text_messages_pass_through(self):
+        messages = [
+            {"role": "system", "content": "You are helpful."},
+            {"role": "user", "content": "Hello"},
+        ]
+        result = parse_vision_messages(messages, llm=None)
+        assert result == messages
 
 
 class TestRemoveSpacesFromEntities: