diff --git a/server-rs/crates/api-server/src/openai_image_generation.rs b/server-rs/crates/api-server/src/openai_image_generation.rs
index 8e94ed52f..35a4b25e5 100644
--- a/server-rs/crates/api-server/src/openai_image_generation.rs
+++ b/server-rs/crates/api-server/src/openai_image_generation.rs
@@ -5,7 +5,7 @@ use platform_image::{
     VECTOR_ENGINE_PROVIDER, build_image_http_client, create_image_edit,
     create_image_edit_with_references, create_image_edit_with_references_and_model,
     create_image_generation_with_model, create_nanobanana_generate_content,
-    missing_reference_images_error,
+    missing_reference_images_error, resolve_image_provider,
 };
 #[cfg(test)]
 use platform_image::{build_image_request_body, images_edit_url, images_generation_url};
@@ -338,13 +338,12 @@ pub(crate) async fn create_openai_image_edit_with_references_and_model(
         "referenceImageCount": reference_images.len(),
     });
     // 中文注释：先做本地参考图校验，避免 provider 客户端缺失时在校验前 panic。
+    // provider 归属与下方非空参考图路径保持一致，按模型解析而不是固定 Tiantoken。
     if reference_images.is_empty() {
+        let provider = resolve_image_provider(model).unwrap_or(ImageProvider::Tiantoken);
         return map_platform_image_result(
             settings,
-            Err(missing_reference_images_error(
-                ImageProvider::Tiantoken,
-                failure_context,
-            )),
+            Err(missing_reference_images_error(provider, failure_context)),
             "image_edit_with_references",
             failure_context,
             request_payload,