diff --git a/funasr/models/fun_asr_nano/inference_vllm.py b/funasr/models/fun_asr_nano/inference_vllm.py
index 331ba90cf..78f567c2d 100644
--- a/funasr/models/fun_asr_nano/inference_vllm.py
+++ b/funasr/models/fun_asr_nano/inference_vllm.py
@@ -39,6 +39,13 @@
 
 dtype_map = {"bf16": torch.bfloat16, "fp16": torch.float16, "fp32": torch.float32}
 
+_LANGUAGE_PROMPT_ALIASES = {
+    "zh": "中文",
+    "en": "英文",
+    "ja": "日文",
+    "ko": "韩文",
+}
+
 
 def _resolve_vllm_dtype(dtype: str) -> str:
     """Use a numerically stable dtype for the Qwen3 language model."""
@@ -462,6 +469,8 @@ def _build_prompt_text(
         itn: bool = True,
     ) -> str:
         """Build the ASR prompt string."""
+        if language:
+            language = _LANGUAGE_PROMPT_ALIASES.get(language.lower(), language)
         hotwords = hotwords or []
         if len(hotwords) > 0:
             hotwords_str = ", ".join(hotwords)
diff --git a/tests/test_fun_asr_nano_vllm_prompt.py b/tests/test_fun_asr_nano_vllm_prompt.py
new file mode 100644
index 000000000..903a8af77
--- /dev/null
+++ b/tests/test_fun_asr_nano_vllm_prompt.py
@@ -0,0 +1,14 @@
+from funasr.models.fun_asr_nano.inference_vllm import FunASRNanoVLLM
+
+
+def test_iso_language_aliases_use_documented_prompt_names():
+    model = object.__new__(FunASRNanoVLLM)
+
+    assert model._build_prompt_text(language="zh") == "语音转写成中文："
+    assert model._build_prompt_text(language="en") == "语音转写成英文："
+
+
+def test_custom_language_prompt_is_preserved():
+    model = object.__new__(FunASRNanoVLLM)
+
+    assert model._build_prompt_text(language="粤语") == "语音转写成粤语："