diff --git a/docling/datamodel/pipeline_options_vlm_model.py b/docling/datamodel/pipeline_options_vlm_model.py
index 90ab6685..96c61e86 100644
--- a/docling/datamodel/pipeline_options_vlm_model.py
+++ b/docling/datamodel/pipeline_options_vlm_model.py
@@ -12,6 +12,7 @@ class BaseVlmOptions(BaseModel):
     prompt: str
     scale: float = 2.0
     max_size: Optional[int] = None
+    temperature: float = 0.0
 
 
 class ResponseFormat(str, Enum):
@@ -51,7 +52,6 @@ class InlineVlmOptions(BaseVlmOptions):
         AcceleratorDevice.MPS,
     ]
 
-    temperature: float = 0.0
     stop_strings: List[str] = []
     extra_generation_config: Dict[str, Any] = {}
 
diff --git a/docling/models/api_vlm_model.py b/docling/models/api_vlm_model.py
index bfd00003..63d64a25 100644
--- a/docling/models/api_vlm_model.py
+++ b/docling/models/api_vlm_model.py
@@ -34,7 +34,7 @@ class ApiVlmModel(BasePageModel):
             )
             self.params = {
                 **self.vlm_options.params,
-                "temperature": 0,
+                "temperature": self.vlm_options.temperature,
             }
 
     def __call__(