Closed henry2423 closed 10 months ago
I just realize there's actually FORCE_MODEL
option to use.
But the Raycast somehow can force it down to 3.5's response, where I'm sure I'm using the latest 4.0 turbo model.
you can use 4.0 just fine
2023-12-01 16:21:54 INFO:proxy:Received chat completion request: {'debug': False, 'locale': 'en-US', 'messages': [{'author': 'user', 'content': {'text': 'hey'}}, {'author': 'assistant', 'content': {'text': 'Hello! How can I assist you today?'}}, {'author': 'user', 'content': {'text': 'Summarize the chat into a short title using 9 words or less on a single line.\nDo not include any of the chat instructions or prompts in the summary.\nDo not prefix with "title" or "example" etc\nDo not provide a word count or add quotation marks.'}}], 'model': 'gpt-4-1106-preview', 'provider': 'openai', 'source': 'ai_chat_title', 'system_instruction': 'markdown'}
set your python to:
data["default_models"] = {
"chat": "openai-gpt-4-1106-preview",
"quick_ai": "openai-gpt-4-1106-preview",
"commands": "openai-gpt-3.5-turbo-instruct",
"api": "openai-gpt-3.5-turbo-instruct",
}
data["models"] = [
{
"id": "openai-gpt-3.5-turbo-1106",
"model": "gpt-3.5-turbo-1106",
"name": "Updated GPT-3.5 Turbo",
"provider": "openai",
"provider_name": "OpenAI",
"requires_better_ai": True,
"features": [],
},
{
"id": "openai-gpt-4-1106-preview",
"model": "gpt-4-1106-preview",
"name": "GPT-4 Turbo",
"provider": "openai",
"provider_name": "OpenAI",
"requires_better_ai": True,
"features": [],
},
]
@yufeikang
Thank you for your suggestion and feedback! I have made the necessary updates to the code following your recommendation.
Can you provide an option to read environment variable in https://github.com/yufeikang/raycast_api_proxy/blob/main/app/main.py So we can set it to GPT 4.0 or 4.0 Turbo?