From ca4e07a40b4ff11989a25d84b517d3d0049f93ae Mon Sep 17 00:00:00 2001
From: Timothy Jaeryang Baek <tim@openwebui.com>
Date: Tue, 25 Aug 2026 11:07:54 -0400
Subject: [PATCH] refac

---
 backend/open_webui/routers/openai.py   | 17 +++++++++++++++--
 backend/open_webui/utils/middleware.py | 13 +++++++++++--
 2 files changed, 26 insertions(+), 4 deletions(-)

diff --git a/backend/open_webui/routers/openai.py b/backend/open_webui/routers/openai.py
index 5988006059..3bc56ebcf6 100644
--- a/backend/open_webui/routers/openai.py
+++ b/backend/open_webui/routers/openai.py
@@ -1359,8 +1359,21 @@ def convert_to_responses_payload(payload: dict) -> dict:
                     content_parts.append({'type': text_type, 'text': part.get('text', '')})
                 elif part.get('type') == 'image_url':
                     url_data = part.get('image_url', {})
-                    url = url_data.get('url', '') if isinstance(url_data, dict) else url_data
-                    content_parts.append({'type': 'input_image', 'image_url': url})
+                    if isinstance(url_data, dict):
+                        url = url_data.get('url', '')
+                        detail = url_data.get('detail') or 'auto'
+                    else:
+                        url = url_data if isinstance(url_data, str) else ''
+                        detail = 'auto'
+                    content_parts.append({'type': 'input_image', 'image_url': url, 'detail': detail})
+                elif part.get('type') == 'file':
+                    # OpenAI-compatible proxy path only. Open WebUI attachments are handled
+                    # separately via metadata.files/RAG and must not be converted here.
+                    file = part.get('file')
+                    if isinstance(file, dict):
+                        file_part = {k: file[k] for k in ('file_id', 'file_data', 'filename') if k in file}
+                        if 'file_id' in file_part or 'file_data' in file_part:
+                            content_parts.append({'type': 'input_file', **file_part})
         else:
             content_parts = [{'type': text_type, 'text': str(content)}]
 
diff --git a/backend/open_webui/utils/middleware.py b/backend/open_webui/utils/middleware.py
index d2dd962641..cc46184d4a 100644
--- a/backend/open_webui/utils/middleware.py
+++ b/backend/open_webui/utils/middleware.py
@@ -2123,7 +2123,13 @@ async def convert_url_images_to_base64(form_data, user=None):
                 new_content.append(item)
                 continue
 
-            image_url = item.get('image_url', {}).get('url', '')
+            image_url_data = item.get('image_url', {})
+            if isinstance(image_url_data, dict):
+                image_url = image_url_data.get('url') or ''
+            elif isinstance(image_url_data, str):
+                image_url = image_url_data
+            else:
+                image_url = ''
             if image_url.startswith('data:image/'):
                 new_content.append(item)
                 continue
@@ -2131,10 +2137,13 @@ async def convert_url_images_to_base64(form_data, user=None):
             try:
                 base64_data = await get_image_base64_from_url(image_url, user=user)
                 if base64_data:
+                    image_url_payload = {'url': base64_data}
+                    if isinstance(image_url_data, dict) and image_url_data.get('detail'):
+                        image_url_payload['detail'] = image_url_data['detail']
                     new_content.append(
                         {
                             'type': 'image_url',
-                            'image_url': {'url': base64_data},
+                            'image_url': image_url_payload,
                         }
                     )
                 else: