Hi, I see a new model called: ChatGPT-4o-latest.
Which is the difference between the classical gpt-4o and gpt-4o-mini?
I see this in the documentation but it is not completely clear. What is the key difference?
Ref: https://platform.openai.com/docs/models/continuous-model-upgrades
I also saw that in https://chat.lmsys.org/ is the current best model…
I just noticed this too!
If I were to take a wild guess I’d say it might be whatever the latest model used by the ChatGPT interface is, something that many have been asking for!
Yes, @trenton.dambrowitz .
I assumed the same but is it currently the high-intelligence model or is gpt-4o?
I think we need some clarification on this.
Thank you for your participation
Initially, this new model appears to exhibit superior reasoning capabilities compared to other “gpt-4o*” models.
Collated everything about this model here:
Where did you see it? Is this only available through the API?
4o-latest points to whatever model they’re using to power the ChatGPT UI, which is neat and helpful.
You’d use this in your code if you do not want to have to continually update for each new model update. It looks as though this is currently pointed at gpt-4o-2024-05-13.
This is super helpful information. Everyone always asks what ChatGPT is pointed at. I didn’t know it was public information.
4o mini is a new model which is generally less-creative than the full model, and it is far more cost effective. It’s best use is for routine, non-higher-reasoning/creative business tasks, like data extraction.
The actual latest model released, but not attached to chatgpt-4o-latest, is gpt-4o-2024-08-06 is a significant update. It allows Structured Outputs, it’s costs have been reduced by 50%, and it’s output has been increased from 4k to 16k tokens.
At least for tha API, chatgpt-4o-latest has a wider (100k) context window compared to gpt-4o (which has 30k)
In summary, this model is used in the ChatGPT interface, is that correct? Or not?
