diff --git a/python/sglang/multimodal_gen/configs/sample/sampling_params.py b/python/sglang/multimodal_gen/configs/sample/sampling_params.py
index e5388df60..fd00f82dc 100644
--- a/python/sglang/multimodal_gen/configs/sample/sampling_params.py
+++ b/python/sglang/multimodal_gen/configs/sample/sampling_params.py
@@ -361,8 +361,6 @@ class SamplingParams:
 
         # TODO: SamplingParams should not rely on ServerArgs
         pipeline_config = server_args.pipeline_config
-        if not isinstance(self.prompt, str):
-            raise TypeError(f"`prompt` must be a string, but got {type(self.prompt)}")
 
         if self.guidance_scale is None:
             try:
diff --git a/python/sglang/multimodal_gen/runtime/entrypoints/diffusion_generator.py b/python/sglang/multimodal_gen/runtime/entrypoints/diffusion_generator.py
index ace021b20..12631f137 100644
--- a/python/sglang/multimodal_gen/runtime/entrypoints/diffusion_generator.py
+++ b/python/sglang/multimodal_gen/runtime/entrypoints/diffusion_generator.py
@@ -8,6 +8,7 @@ This module provides a consolidated interface for generating images/videos using
 diffusion models.
 """
 
+import dataclasses
 import multiprocessing as mp
 import os
 import time
@@ -165,15 +166,21 @@ class DiffGenerator:
         """
         # 1. prepare requests
         prompts = self._resolve_prompts(sampling_params_kwargs.get("prompt"))
-        sampling_params = SamplingParams.from_user_sampling_params_args(
+        sampling_params_orig = SamplingParams.from_user_sampling_params_args(
             self.server_args.model_path,
             server_args=self.server_args,
             **sampling_params_kwargs,
         )
+        user_output_file_name = sampling_params_kwargs.get("output_file_name")
 
         requests: list[Req] = []
         for p in prompts:
-            sampling_params.prompt = p
+            sampling_params = dataclasses.replace(
+                sampling_params_orig,
+                prompt=p,
+                output_file_name=user_output_file_name,
+            )
+            sampling_params._set_output_file_name()
             req = prepare_request(
                 server_args=self.server_args,
                 sampling_params=sampling_params,
diff --git a/python/sglang/multimodal_gen/runtime/entrypoints/utils.py b/python/sglang/multimodal_gen/runtime/entrypoints/utils.py
index 95bc98ef7..3d9d8b45b 100644
--- a/python/sglang/multimodal_gen/runtime/entrypoints/utils.py
+++ b/python/sglang/multimodal_gen/runtime/entrypoints/utils.py
@@ -298,6 +298,9 @@ def prepare_request(
 
     req.adjust_size(server_args)
 
+    if not isinstance(req.prompt, str):
+        raise TypeError(f"`prompt` must be a string, but got {type(req.prompt)}")
+
     if (req.width is not None and req.width <= 0) or (
         req.height is not None and req.height <= 0
     ):