diff --git a/metagpt/provider/base_llm.py b/metagpt/provider/base_llm.py
index 80e51f8ac..b3aa1b340 100644
--- a/metagpt/provider/base_llm.py
+++ b/metagpt/provider/base_llm.py
@@ -215,7 +215,10 @@ class BaseLLM(ABC):
 
     def get_choice_text(self, rsp: dict) -> str:
         """Required to provide the first text of choice"""
-        return rsp.get("choices")[0]["message"]["content"]
+        message = rsp.get("choices")[0]["message"]
+        if "reasoning_content" in message:
+            self.reasoning_content = message["reasoning_content"]
+        return message["content"]
 
     def get_choice_delta_text(self, rsp: dict) -> str:
         """Required to provide the first text of stream choice"""
diff --git a/metagpt/provider/openai_api.py b/metagpt/provider/openai_api.py
index bd9c02231..5e718f45f 100644
--- a/metagpt/provider/openai_api.py
+++ b/metagpt/provider/openai_api.py
@@ -91,8 +91,12 @@ class OpenAILLM(BaseLLM):
         )
         usage = None
         collected_messages = []
+        collected_reasoning_messages = []
         has_finished = False
         async for chunk in response:
+            if hasattr(chunk.choices[0].delta, "reasoning_content"):
+                collected_reasoning_messages.append(chunk.choices[0].delta.reasoning_content)  # for deepseek
+                continue
             chunk_message = chunk.choices[0].delta.content or "" if chunk.choices else ""  # extract the message
             finish_reason = (
                 chunk.choices[0].finish_reason if chunk.choices and hasattr(chunk.choices[0], "finish_reason") else None
@@ -118,6 +122,8 @@ class OpenAILLM(BaseLLM):
 
         log_llm_stream("\n")
         full_reply_content = "".join(collected_messages)
+        if collected_reasoning_messages:
+            self.reasoning_content = "".join(collected_reasoning_messages)
         if not usage:
             # Some services do not provide the usage attribute, such as OpenAI or OpenLLM
             usage = self._calc_usage(messages, full_reply_content)