Merge pull request #448 from sudoleg/feat/gpt-5

GPT-5 models

Author: sudoleg

README.md

@@ -23,8 +23,8 @@ YouTubeGPT lets you **summarize and chat (Q&A)** with YouTube videos. Its featur
 
 ### :robot: Choose from different OpenAI models
 
-- currently available: gpt-3.5-turbo, gpt-4 (turbo), gpt-4o (mini)
-- by choosing a different model, you can summarize even longer videos and potentially get better responses
+- currently available: ChatGPT 4-5 (incl. nano & mini) and *continuously updated* with new models
+- by choosing a different model, you can summarize even longer videos and get better responses
 
 ### :gear: Experiment with settings
 

config.json

@@ -7,7 +7,7 @@
     },
     "default_model": {
         "embeddings": "text-embedding-3-small",
-        "gpt": "gpt-4o-mini"
+        "gpt": "gpt-4.1-nano"
     },
     "available_models": {
         "embeddings": [
@@ -17,6 +17,11 @@
         ],
         "gpts": [
             "gpt-4.1-nano",
+            "gpt-5.1-chat-latest",
+            "gpt-5.1",
+            "gpt-5-nano",
+            "gpt-5-mini",
+            "gpt-5",
             "gpt-4.1-mini",
             "gpt-4.1",
             "gpt-4o-mini",

modules/summary.py

@@ -1,7 +1,7 @@
 import logging
 
-from langchain.chat_models import BaseChatModel
 from langchain.messages import HumanMessage, SystemMessage
+from langchain_openai import ChatOpenAI
 
 from .helpers import num_tokens_from_string, read_file
 
@@ -19,6 +19,11 @@
     "gpt-4.1-nano": {"total": 1047576, "output": 32768},
     "gpt-4.1-mini": {"total": 1047576, "output": 32768},
     "gpt-4.1": {"total": 1047576, "output": 32768},
+    "gpt-5.1-chat-latest": {"total": 400000, "output": 128000},
+    "gpt-5.1": {"total": 400000, "output": 128000},
+    "gpt-5-nano": {"total": 400000, "output": 128000},
+    "gpt-5-mini": {"total": 400000, "output": 128000},
+    "gpt-5": {"total": 400000, "output": 128000},
 }
 
 
@@ -35,13 +40,13 @@ def log_error(self):
         logging.error("Transcript too long for %s.", self.model_name, exc_info=True)
 
 
-def get_transcript_summary(transcript_text: str, llm: BaseChatModel, **kwargs):
+def get_transcript_summary(transcript_text: str, llm: ChatOpenAI, **kwargs):
     """
     Generates a summary from a video transcript using a language model.
 
     Args:
         transcript_text (str): The full transcript text of the video.
-        llm (BaseChatModel): The language model instance to use for generating the summary.
+        llm (ChatOpenAI): The language model instance to use for generating the summary.
         **kwargs: Optional keyword arguments.
             - custom_prompt (str): A custom prompt to replace the default summary request.
 

modules/ui.py

@@ -95,7 +95,7 @@ def display_model_settings_sidebar():
             )
         if model != get_default_config_value("default_model.gpt"):
             st.warning(
-                """:warning: Be aware of the higher costs and potentially higher latencies when using more advanced models (like gpt-4 and gpt-4o). You can see details (incl. costs) about the models and compare them [here](https://platform.openai.com/docs/models/compare)."""
+                """:warning: Be aware of the higher costs and latencies when using more advanced (reasoning) models (like gpt-5). You can see details (incl. costs) about the models and compare them [here](https://platform.openai.com/docs/models/compare)."""
             )
 
 

pages/chat.py

@@ -148,7 +148,7 @@ def save_response_to_lib():
         temperature=st.session_state.temperature,
         model=st.session_state.model,
         top_p=st.session_state.top_p,
-        max_tokens=2048,
+        # max_tokens=2048,
     )
     openai_embedding_model = OpenAIEmbeddings(
         api_key=st.session_state.openai_api_key,

pages/summary.py

@@ -116,8 +116,7 @@ def save_summary_to_lib():
                     temperature=st.session_state.temperature,
                     model=st.session_state.model,
                     top_p=st.session_state.top_p,
-                    max_tokens=2048,
-                    use_responses_api=False,
+                    # max_completion_tokens=4096,
                 )
                 with st.spinner("Summarizing video :gear: Hang on..."):
                     if custom_prompt: