Spaces:

KingNish

/

OpenGPT-4o

like 2.13k

Running on Zero

this is not chtgpt 4, he is faking the AI behind the space to get likes

The AI is referred to as OpenGPT 4, an open-source variant of ChatGPT 4 (Plus).
It boasts several features not available in the ChatGPT 4 (Free Tier), such as:

1. Image Generation
2. Voice Chat
3. Video Chat

This space is using the llm Mixtral-8x7B-Instruct-v0.1, a model far less capable than ChatGPT4 when it comes to text generation and chatting, in combination with several smaller models. This is different from gpt4o, which is advertised as being "natively" multimodal, therefore presumably being a single model. That doesn't mean it is a bad project, just that the title of the space "KingNish/GPT-4o" is misleading, as this is not gpt4o, and not even close due to the points mentioned previously. It would be great if you could rename it to something like "OpenGPTOmni", without making use of the "GPT4o" name, as it is misleading to users that don't know the underlying models used and assume it is GPT4o as advertised by OpenAI.

@nlpguy Well its name on both main page and demo page is OpenGPT 4o

As you mentioned, the name in the URL is GPT-4o, which I have now changed. I hope this eliminates any confusion.

Thanks :)

This space is using the llm Mixtral-8x7B-Instruct-v0.1, a model far less capable than ChatGPT4 when it comes to text generation and chatting, in combination with several smaller models. This is different from gpt4o, which is advertised as being "natively" multimodal, therefore presumably being a single model. That doesn't mean it is a bad project, just that the title of the space "KingNish/GPT-4o" is misleading, as this is not gpt4o, and not even close due to the points mentioned previously. It would be great if you could rename it to something like "OpenGPTOmni", without making use of the "GPT4o" name, as it is misleading to users that don't know the underlying models used and assume it is GPT4o as advertised by OpenAI.

fyi: its not using mixtral, its using idefics2 8b for the chat, mixtral for the voice I think

Why not use llama-3-70b for voice chat?

Why not use llama-3-70b for voice chat?

Its size is very big, which make it slow