What would it matter at that point though? You could nail the leaker to the wall and sue everyone who posts it online, but once it’s out on the internet, people will still be using it.
This looks more promising to me than what I saw yesterday. There is an "Open" aspect to their objectives and they do say:
> We’re committed to empower the AI community with open technology. Our open models sets the bar for efficiency, and are available for free, with fully permissive license.
(that's something people were looking for yesterday)
It's a shame they couldn't find another business model. I've found a lot of value in Mistral, but none of the raw models were what worked for me. The models I've been using were derived from mistral... but it was the open source community that made it useful.
Did they try to find another business model or just take a bribe from Microsoft to close down their models? Their pricing (when factoring in tokenization) is more or less the same as GPT-4-turbo for worse models. This seems like a pretty clear cut case of market manipulation and I’d be surprised if they don’t run afoul of EU regulators, since they also were the beneficiaries of various forms of state support.
