diff --git a/server-rs/crates/platform-image/src/image_provider/runtime/executor.rs b/server-rs/crates/platform-image/src/image_provider/runtime/executor.rs
index fa86a2753..5b4926924 100644
--- a/server-rs/crates/platform-image/src/image_provider/runtime/executor.rs
+++ b/server-rs/crates/platform-image/src/image_provider/runtime/executor.rs
@@ -216,6 +216,7 @@ pub async fn create_image_generation_with_model(
             "ImageProvider 图片生成 HTTP 返回"
         );
         let response_text = response.body;
+        let task_id_prefix = image_task_id_prefix(settings.provider, None);
         match handle_image_response(
             http_client,
             request_url.as_str(),
@@ -228,7 +229,7 @@ pub async fn create_image_generation_with_model(
             Some(prompt.chars().count()),
             Some(reference_images.len()),
             candidate_count,
-            "vector-engine",
+            task_id_prefix.as_str(),
             settings.request_deadline,
         )
         .await
@@ -406,6 +407,7 @@ pub async fn create_nanobanana_generate_content(
         "ImageProvider nanobanana2 图片生成 HTTP 返回"
     );
     let response_text = response.body;
+    let task_id_prefix = image_task_id_prefix(settings.provider, Some("nanobanana"));
     handle_image_response(
         http_client,
         request_url.as_str(),
@@ -418,7 +420,7 @@ pub async fn create_nanobanana_generate_content(
         Some(prompt.chars().count()),
         Some(reference_image_count),
         1,
-        "vector-engine-nanobanana",
+        task_id_prefix.as_str(),
         settings.request_deadline,
     )
     .await
@@ -657,6 +659,7 @@ pub async fn create_image_edit_with_references_and_model(
             "ImageProvider 图片编辑 HTTP 返回"
         );
         let response_text = response.body;
+        let task_id_prefix = image_task_id_prefix(settings.provider, Some("edit"));
         match handle_image_response(
             http_client,
             request_url.as_str(),
@@ -669,7 +672,7 @@ pub async fn create_image_edit_with_references_and_model(
             Some(prompt.chars().count()),
             Some(reference_image_count),
             candidate_count,
-            "vector-engine-edit",
+            task_id_prefix.as_str(),
             settings.request_deadline,
         )
         .await
@@ -695,7 +698,7 @@ pub async fn create_image_edit_with_references_and_model(
 /// executor 用 `settings.base_url`/`settings.api_key` 发请求，如果调用方传入了属于
 /// 另一个 provider 的模型，就会把错误的密钥/网关当成目标端点。这里显式 fail fast，
 /// 避免静默地把请求发到不匹配的 provider。
-fn ensure_provider_matches_model(
+pub(crate) fn ensure_provider_matches_model(
     settings: &ImageProviderSettings,
     model: &str,
     failure_context: &str,
@@ -718,6 +721,15 @@ fn ensure_provider_matches_model(
     Ok(())
 }
 
+/// 兜底 task id 前缀：必须跟随实际 provider，否则 Tiantoken 任务的
+/// `task_id` 会带着 `vector-engine` 前缀进入持久化与审计。
+fn image_task_id_prefix(provider: ImageProvider, task_kind: Option<&str>) -> String {
+    match task_kind {
+        Some(kind) => format!("{}-{kind}", provider.as_str()),
+        None => provider.as_str().to_string(),
+    }
+}
+
 fn preferred_image_upstream_model(requested_model: &str) -> &str {
     // Provider routing is selected by api-server at the task boundary. Keep the
     // concrete value intact here so retries stay on the same model and legacy