Replace 3.5 turbo with new 4o mini model released this week

bot/bot.py

@@ -94,7 +94,7 @@ async def register_user_if_not_exists(update: Update, context: CallbackContext,
     n_used_tokens = db.get_user_attribute(user.id, "n_used_tokens")
     if isinstance(n_used_tokens, int) or isinstance(n_used_tokens, float):  # old format
         new_n_used_tokens = {
-            "gpt-3.5-turbo": {
+            "gpt-4o-mini": {
                 "n_input_tokens": 0,
                 "n_output_tokens": n_used_tokens
             }
@@ -185,7 +185,7 @@ async def _vision_message_handle_fn(
     user_id = update.message.from_user.id
     current_model = db.get_user_attribute(user_id, "current_model")
 
-    if current_model != "gpt-4-vision-preview" and current_model != "gpt-4o":
+    if current_model not in ["gpt-4-vision-preview", "gpt-4o", "gpt-4o-mini"]:
         await update.message.reply_text(
             "🥲 Images processing is only available for <b>gpt-4-vision-preview</b> and <b>gpt-4o</b> model. Please change your settings in /settings",
             parse_mode=ParseMode.HTML,
@@ -566,7 +566,7 @@ async def new_dialog_handle(update: Update, context: CallbackContext):
 
     user_id = update.message.from_user.id
     db.set_user_attribute(user_id, "last_interaction", datetime.now())
-    db.set_user_attribute(user_id, "current_model", "gpt-3.5-turbo")
+    db.set_user_attribute(user_id, "current_model", "gpt-4o-mini")
 
     db.start_new_dialog(user_id)
     await update.message.reply_text("Starting new dialog ✅")

bot/openai_utils.py

@@ -25,8 +25,8 @@
 
 
 class ChatGPT:
-    def __init__(self, model="gpt-3.5-turbo"):
-        assert model in {"text-davinci-003", "gpt-3.5-turbo-16k", "gpt-3.5-turbo", "gpt-4", "gpt-4o", "gpt-4-1106-preview", "gpt-4-vision-preview"}, f"Unknown model: {model}"
+    def __init__(self, model="gpt-4o-mini"):
+        assert model in {"text-davinci-003", "gpt-3.5-turbo-16k", "gpt-4o-mini", "gpt-4", "gpt-4o", "gpt-4-1106-preview", "gpt-4-vision-preview"}, f"Unknown model: {model}"
         self.model = model
 
     async def send_message(self, message, dialog_messages=[], chat_mode="assistant"):
@@ -37,7 +37,7 @@ async def send_message(self, message, dialog_messages=[], chat_mode="assistant")
         answer = None
         while answer is None:
             try:
-                if self.model in {"gpt-3.5-turbo-16k", "gpt-3.5-turbo", "gpt-4", "gpt-4o", "gpt-4-1106-preview", "gpt-4-vision-preview"}:
+                if self.model in {"gpt-3.5-turbo-16k", "gpt-4o-mini", "gpt-4", "gpt-4o", "gpt-4-1106-preview", "gpt-4-vision-preview"}:
                     messages = self._generate_prompt_messages(message, dialog_messages, chat_mode)
 
                     r = await openai.ChatCompletion.acreate(
@@ -78,7 +78,7 @@ async def send_message_stream(self, message, dialog_messages=[], chat_mode="assi
         answer = None
         while answer is None:
             try:
-                if self.model in {"gpt-3.5-turbo-16k", "gpt-3.5-turbo", "gpt-4","gpt-4o", "gpt-4-1106-preview"}:
+                if self.model in {"gpt-3.5-turbo-16k", "gpt-4o-mini", "gpt-4","gpt-4o", "gpt-4-1106-preview"}:
                     messages = self._generate_prompt_messages(message, dialog_messages, chat_mode)
 
                     r_gen = await openai.ChatCompletion.acreate(
@@ -186,7 +186,7 @@ async def send_vision_message_stream(
         answer = None
         while answer is None:
             try:
-                if self.model == "gpt-4-vision-preview" or self.model == "gpt-4o":
+                if self.model in ["gpt-4-vision-preview", "gpt-4o", "gpt-4o-mini"]:
                     messages = self._generate_prompt_messages(
                         message, dialog_messages, chat_mode, image_buffer
                     )
@@ -289,13 +289,13 @@ def _postprocess_answer(self, answer):
         answer = answer.strip()
         return answer
 
-    def _count_tokens_from_messages(self, messages, answer, model="gpt-3.5-turbo"):
+    def _count_tokens_from_messages(self, messages, answer, model="gpt-4o-mini"):
         encoding = tiktoken.encoding_for_model(model)
 
         if model == "gpt-3.5-turbo-16k":
             tokens_per_message = 4  # every message follows <im_start>{role/name}\n{content}<im_end>\n
             tokens_per_name = -1  # if there's a name, the role is omitted
-        elif model == "gpt-3.5-turbo":
+        elif model == "gpt-4o-mini":
             tokens_per_message = 4
             tokens_per_name = -1
         elif model == "gpt-4":

config/models.yml

@@ -1,16 +1,38 @@
-available_text_models: ["gpt-3.5-turbo", "gpt-3.5-turbo-16k", "gpt-4-1106-preview", "gpt-4-vision-preview", "gpt-4", "text-davinci-003", "gpt-4o"]
+available_text_models:
+  [
+    "gpt-4o-mini",
+    "gpt-3.5-turbo-16k",
+    "gpt-4-1106-preview",
+    "gpt-4-vision-preview",
+    "gpt-4",
+    "text-davinci-003",
+    "gpt-4o",
+  ]
 
 info:
+  gpt-4o-mini:
+    type: chat_completion
+    name: GPT-4O-MINI
+    description: ChatGPT 4o mini. It's <b>fast</b> and <b>cheap</b>. Ideal for everyday tasks.
+
+    price_per_1000_input_tokens: 0.00015
+    price_per_1000_output_tokens: 0.0006
+
+    scores:
+      Smart: 5
+      Fast: 5
+      Cheap: 5
+
   gpt-3.5-turbo:
     type: chat_completion
-    name: ChatGPT
-    description: ChatGPT is that well-known model. It's <b>fast</b> and <b>cheap</b>. Ideal for everyday tasks. If there are some tasks it can't handle, try the <b>GPT-4</b>.
+    name: GPT-3.5 Turbo
+    description: Old. Dont use this.
 
-    price_per_1000_input_tokens: 0.0015
-    price_per_1000_output_tokens: 0.002
+    price_per_1000_input_tokens: 0.003
+    price_per_1000_output_tokens: 0.004
 
     scores:
-      Smart: 3
+      Smart: 1
       Fast: 5
       Cheap: 5
 
@@ -92,8 +114,8 @@ info:
       Cheap: 3
 
   dalle-2:
-     type: image
-     price_per_1_image: 0.018  # 512x512
+    type: image
+    price_per_1_image: 0.018 # 512x512
 
   whisper:
     type: audio