diff --git a/README.md b/README.md
index ae65c7a..49bbc10 100644
--- a/README.md
+++ b/README.md
@@ -77,6 +77,12 @@ conversation_list = chatbot.get_conversation_list()
 # Get the available models (not hardcore)
 models = chatbot.get_available_llm_models()
 
+# Get image link.
+# Work only for model "CohereForAI/c4ai-command-r-plus"
+chat_result = chatbot.chat("Draw a cat.")
+print(chat_result.get_final_text())
+print(chat_result.get_image_link())
+
 # Switch model with given index
 chatbot.switch_llm(0) # Switch to the first model
 chatbot.switch_llm(1) # Switch to the second model
diff --git a/src/hugchat/hugchat.py b/src/hugchat/hugchat.py
index d1d8016..0cacfa7 100644
--- a/src/hugchat/hugchat.py
+++ b/src/hugchat/hugchat.py
@@ -711,6 +711,7 @@ def _stream_query(
             resp.encoding = 'utf-8'
 
             if resp.status_code != 200:
+                
                 retry_count -= 1
                 if retry_count <= 0:
                     raise exceptions.ChatError(
@@ -725,6 +726,12 @@ def _stream_query(
                     if obj.__contains__("type"):
                         _type = obj["type"]
 
+                        if _type == "file":
+                            _sha = obj["sha"]
+                            _image_link = f"{self.hf_base_url}/chat/conversation/{conversation}/output/{_sha}"
+                            yield {"type": _type, "image_link": _image_link}
+                            continue
+
                         if _type == "finalAnswer":
                             final_answer = obj
                             break_flag = True
diff --git a/src/hugchat/message.py b/src/hugchat/message.py
index b26caf0..b4e962c 100644
--- a/src/hugchat/message.py
+++ b/src/hugchat/message.py
@@ -7,6 +7,7 @@
 RESPONSE_TYPE_STREAM = "stream"
 RESPONSE_TYPE_WEB = "webSearch"
 RESPONSE_TYPE_STATUS = "status"
+RESPONSE_TYPE_IMAGE = "file"
 MSGTYPE_ERROR = "error"
 
 MSGSTATUS_PENDING = 0
@@ -73,6 +74,7 @@ def __init__(
         self.g = g
         self._stream_yield_all = _stream_yield_all
         self.web_search = web_search
+        self.image_link = None
 
     @property
     def text(self) -> str:
@@ -127,6 +129,9 @@ def __next__(self) -> dict:
                     self.web_search_done = True
                 elif t == RESPONSE_TYPE_STATUS:
                     pass
+            elif t == RESPONSE_TYPE_IMAGE:
+                if not a.__contains__("sha"):
+                    self.image_link =  a["image_link"]   
             else:
                 if "Model is overloaded" in str(a):
                     self.error = ModelOverloadedError(
@@ -185,6 +190,13 @@ def get_search_sources(self) -> list:
         """
         return self.web_search_sources
 
+    def get_image_link(self) -> str:
+        """
+        :Return:
+            - self.image_link
+        """
+        return self.image_link
+
     def search_enabled(self) -> bool:
         """
         :Return: