gpt4o debug

app.py

@@ -160,7 +160,7 @@ def reasoner(query: str, documents: list[str], llm_client: Any):
             message = "Using Claude Sonnet to reason and verify the claim..."
         elif selected_reasoner == "GPT-4o":
             message = "Using GPT-4o to analyze and verify the claim in detail..."
-        else:
+        elif selected_reasoner == "o3-mini":
             message = "Using o3-mini to quickly analyze the claim..."
 
         if not documents or len(documents) == 0:
@@ -215,7 +215,13 @@ if prompt := st.chat_input("Type here"):
         api_key = os.getenv("openai_key")
         options["API_KEY"] = api_key
         options["model_family"] = "OpenAI"
-        options["model_name"] = "gpt-4o-2024-05-13"
+        options["model_name"] = "gpt-4o-2024-11-20"
+
+    elif selected_reasoner == "o3-mini":
+        api_key = os.getenv("openai_key")
+        options["API_KEY"] = api_key
+        options["model_family"] = "OpenAI"
+        options["model_name"] = "o3-mini-2025-01-31"
 
     llm_client = LLMReasoner(options)
 

llm_reasoner.py

@@ -30,21 +30,28 @@ class LLMReasoner():
 
     def make_openai_chat_completions_api_call(self, prompt):
         try:
-            response = self.client.chat.completions.create(
-                model=self.model_name,
-                messages=prompt,
-                temperature=self.temp,
-                max_completion_tokens=self.max_tokens,
-                top_p=self.top_p,
-                frequency_penalty=self.frequency_penalty,
-                presence_penalty=self.presence_penalty
-            )
+            if "gpt-4o" in self.model_name:
+                response = self.client.chat.completions.create(
+                    model=self.model_name,
+                    messages=prompt,
+                    temperature=self.temp,
+                    max_completion_tokens=self.max_tokens,
+                    top_p=self.top_p,
+                    frequency_penalty=self.frequency_penalty,
+                    presence_penalty=self.presence_penalty
+                )
+            elif "o3-mini" in self.model_name:
+                response = self.client.chat.completions.create(
+                    model=self.model_name,
+                    messages=prompt,
+                    reasoning_effort="medium"
+                )
             return self.parse_chat_completions_api_response(response)
         except openai.APIConnectionError as e:
             print("The server could not be reached")
             print(e.__cause__)  # an underlying Exception, likely raised within httpx.
             time.sleep(60)
-            return self.make_openai_api_call(prompt)
+            return self.make_openai_chat_completions_api_call(prompt)
         except openai.RateLimitError as e:
             print("Rate limit error hit")
             exit()
@@ -56,7 +63,7 @@ class LLMReasoner():
             print(e.status_code)
             # print(e.response.data)
             time.sleep(60)
-            return self.make_openai_api_call(prompt)
+            return self.make_openai_chat_completions_api_call(prompt)
 
     def parse_chat_completions_api_response(self, response):
         # print(response.model_dump())