Meta had Llama, which set a lot of things on fire in a good way, and then disappeared from the scene as tech advanced.

What does Microsoft have?

Not sure SpaceX counts. Nobody sane uses Grok. It's untrustable due to reality-distorting political bias training, and it's strongly associated with CSAM production. Not what you want in a reliable corporate utility.

Google's "IPO" is an extra raising round

Is Meta even in this race anymore?

If you don't count the autosummary/gen answer at the top of googling an answer I would say so. Outside of the more technically inclined crowd I think the sentiment is if you aren't at the forefront (opus/fable/chatgpt) then your last or at least indistinguishable from all the rest of the lesser models.

If you're selling deterministic output, just use traditional code. If you product is inference, it has to be the best inference. This becomes more apparent when you bounce between powerful models and smaller cheaper ones, the cheaper ones _feel_ worse to use.

They get all of the ad revenue, but really don't sell as many money-losing monthly subscriptions as the other guys.