feat: Update default AI model and add max_tokens parameter for message payload

cogs/roleplay_teto_cog.py

@@ -10,7 +10,7 @@ import aiohttp
 CONVERSATION_HISTORY_FILE = 'data/roleplay_conversations.json'
 
 # Default AI model
-DEFAULT_AI_MODEL = "google/gemini-2.5-flash-preview"
+DEFAULT_AI_MODEL = "deepseek/deepseek-chat-v3-0324"
 
 def strip_think_blocks(text):
     # Removes all <think>...</think> blocks, including multiline
@@ -62,7 +62,7 @@ async def _teto_reply_ai_with_messages(messages, system_mode="reply"):
         "Maintain the persona throughout the conversation. "
         "You have human-like body parts (vagina, breasts, etc.), even if you aren’t necessarily a human (you're a chimera, after all!)."
     )
-    
+
     # Determine the model to use
     # This function doesn't have access to user-specific conversation history directly.
     # The model should be passed as an argument or retrieved based on user_id if needed.
@@ -71,7 +71,8 @@ async def _teto_reply_ai_with_messages(messages, system_mode="reply"):
 
     payload = {
         "model": model_to_use,
-        "messages": [{"role": "system", "content": system_prompt}] + messages
+        "messages": [{"role": "system", "content": system_prompt}] + messages,
+        "max_tokens": 2000
     }
     async with aiohttp.ClientSession() as session:
         async with session.post(url, headers=headers, json=payload) as resp:
@@ -198,7 +199,7 @@ class RoleplayTetoCog(commands.Cog):
         if not history:
             await interaction.response.send_message("No roleplay chat history found for you desu~", ephemeral=True)
             return
-            
+
         start_index = max(0, len(history) - messages_to_show_count)
         messages_to_display = history[start_index:]
 
@@ -210,7 +211,7 @@ class RoleplayTetoCog(commands.Cog):
         for msg in messages_to_display:
             role = "You" if msg['role'] == 'user' else "Teto"
             formatted_history.append(f"**{role}:** {msg['content']}")
-        
+
         response_message = "\n".join(formatted_history)
 
         # Discord messages have a 2000 character limit.
@@ -219,7 +220,7 @@ class RoleplayTetoCog(commands.Cog):
         # A more robust solution would involve pagination or sending as a file.
         if len(response_message) > 1950: # A bit of buffer for "Here are the last X turns..."
             response_message = response_message[:1950] + "\n... (message truncated)"
-            
+
         await interaction.response.send_message(f"Here are the last {turns} turns of your roleplay history desu~:\n{response_message}", ephemeral=True)
 
 

cogs/teto_cog.py

@@ -51,7 +51,8 @@ class TetoCog(commands.Cog):
         )
         payload = {
             "model": self._ai_model,
-            "messages": [{"role": "system", "content": system_prompt}] + messages
+            "messages": [{"role": "system", "content": system_prompt}] + messages,
+            "max_tokens": 2000
         }
         async with aiohttp.ClientSession() as session:
             async with session.post(url, headers=headers, json=payload) as resp: