The collective game of still just playing catch-up to GPT4, which was released a year ago, while having apparently no special sauce and full well knowing that OpenAI could come up with something much better at any point must be really exhausting.
GPT-4 is an enormous model that took an enormous amount of training. The big news is that smaller teams are getting close to its performance on a small model that can run on a single GPU. No doubt many of these innovations could be scaled up, but pretty much only Google and Microsoft have the compute resources for the behemoth models(Not just the training, but the giant resources required to run the inference for hundreds of millions of users. Google already claims to have surpassed GPT-4 with their unreleased Gemini Ultra model. No doubt OpenAI/Microsoft is sitting on GPT-5 refining it just waiting to leapfrog the competition.
Don't forget about Meta
Also, GPT-4 isn’t actually a model per se. It’s a black-box product that uses a model.
Maybe if OpenAI also seemed like it was getting stronger, especially organizationally. But if I were Mistral and following this quickly while OAI was tripping over its own shoelaces… that’s gotta be very exciting.
Is OpenAI tripping over its own shoelaces? They haven't released GPT-5 but then there was over two years between training GPT-3 and GPT-4 and they spent 9 months safety testing GPT-4 before release.
Commenter is referring to firing and rehiring of sam altman, and the business structure itself having contradictions e.g. nonprofit vs for profit motives and mission
It's more exhausting seeing cheering on a single product provider just because they were first in the age of complaining about giant tech monopolies. Let them cook. The more options, the better. It's only a matter of time before people give OpenAI the Google and Apple treatment.
> "...it appears that not only is Mistral training a version of this so-called “Miqu” model that approaches GPT-4 level performance, but it may, in fact, match or exceed it, if his comments are to be interpreted generously."