diff --git a/docs/offline_inference.md b/docs/offline_inference.md
index 5e8446f01..0c5b10de3 100644
--- a/docs/offline_inference.md
+++ b/docs/offline_inference.md
@@ -198,6 +198,8 @@ For ``LLM`` configuration, refer to [Parameter Documentation](parameters.md).
 * finished(bool): Completion status
 * metrics(fastdeploy.engine.request.RequestMetrics): Performance metrics
 * num_cached_tokens(int): Cached token count (only valid when enable_prefix_caching``` is enabled)
+* num_input_image_tokens(int): Number of input image tokens.
+* num_input_video_tokens(int): Number of input video tokens.
 * error_code(int): Error code
 * error_msg(str): Error message
 
diff --git a/docs/online_serving/README.md b/docs/online_serving/README.md
index 8423530ff..526573a63 100644
--- a/docs/online_serving/README.md
+++ b/docs/online_serving/README.md
@@ -238,6 +238,19 @@ ChatMessage:
     completion_token_ids: Optional[List[int]] = None
     prompt_tokens: Optional[str] = None
     completion_tokens: Optional[str] = None
+UsageInfo:
+    prompt_tokens: int = 0
+    total_tokens: int = 0
+    completion_tokens: Optional[int] = 0
+    prompt_tokens_details: Optional[PromptTokenUsageInfo] = None
+    completion_tokens_details: Optional[CompletionTokenUsageInfo] = None
+PromptTokenUsageInfo:
+    cached_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
+    video_tokens: Optional[int] = None
+CompletionTokenUsageInfo:
+    reasoning_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
 ToolCall:
     id: str = None
     type: Literal["function"] = "function"
@@ -414,6 +427,19 @@ CompletionResponseChoice:
     reasoning_content: Optional[str] = None
     finish_reason: Optional[Literal["stop", "length", "tool_calls"]]
     tool_calls: Optional[List[DeltaToolCall | ToolCall]] = None
+UsageInfo:
+    prompt_tokens: int = 0
+    total_tokens: int = 0
+    completion_tokens: Optional[int] = 0
+    prompt_tokens_details: Optional[PromptTokenUsageInfo] = None
+    completion_tokens_details: Optional[CompletionTokenUsageInfo] = None
+PromptTokenUsageInfo:
+    cached_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
+    video_tokens: Optional[int] = None
+CompletionTokenUsageInfo:
+    reasoning_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
 ToolCall:
     id: str = None
     type: Literal["function"] = "function"
diff --git a/docs/zh/offline_inference.md b/docs/zh/offline_inference.md
index 45e28efb1..bf5296b2d 100644
--- a/docs/zh/offline_inference.md
+++ b/docs/zh/offline_inference.md
@@ -198,6 +198,8 @@ for output in outputs:
 * finished(bool)：标识当前query 是否推理结束
 * metrics(fastdeploy.engine.request.RequestMetrics)：记录推理耗时指标
 * num_cached_tokens(int): 缓存的token数量, 仅在开启 ``enable_prefix_caching``时有效
+* num_input_image_tokens(int): 输入图片token的数量
+* num_input_video_tokens(int): 输入视频token的数量
 * error_code(int): 错误码
 * error_msg(str): 错误信息
 
diff --git a/docs/zh/online_serving/README.md b/docs/zh/online_serving/README.md
index 587a0a9fa..04c5a6868 100644
--- a/docs/zh/online_serving/README.md
+++ b/docs/zh/online_serving/README.md
@@ -237,6 +237,19 @@ ChatMessage:
     completion_token_ids: Optional[List[int]] = None
     prompt_tokens: Optional[str] = None
     completion_tokens: Optional[str] = None
+UsageInfo:
+    prompt_tokens: int = 0
+    total_tokens: int = 0
+    completion_tokens: Optional[int] = 0
+    prompt_tokens_details: Optional[PromptTokenUsageInfo] = None
+    completion_tokens_details: Optional[CompletionTokenUsageInfo] = None
+PromptTokenUsageInfo:
+    cached_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
+    video_tokens: Optional[int] = None
+CompletionTokenUsageInfo:
+    reasoning_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
 ToolCall:
     id: str = None
     type: Literal["function"] = "function"
@@ -410,6 +423,19 @@ CompletionResponseChoice:
     reasoning_content: Optional[str] = None
     finish_reason: Optional[Literal["stop", "length", "tool_calls"]]
     tool_calls: Optional[List[DeltaToolCall | ToolCall]] = None
+UsageInfo:
+    prompt_tokens: int = 0
+    total_tokens: int = 0
+    completion_tokens: Optional[int] = 0
+    prompt_tokens_details: Optional[PromptTokenUsageInfo] = None
+    completion_tokens_details: Optional[CompletionTokenUsageInfo] = None
+PromptTokenUsageInfo:
+    cached_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
+    video_tokens: Optional[int] = None
+CompletionTokenUsageInfo:
+    reasoning_tokens: Optional[int] = None
+    image_tokens: Optional[int] = None
 ToolCall:
     id: str = None
     type: Literal["function"] = "function"