News for hackers. Stuff that matters.
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by verdverm on Tuesday July 28, 2026 @11:00AM (#49084979)
Google is apparently taking a different stance and offering distillation as a paid product

https://docs.cloud.google.com/gemini-enterprise-agent-platfo...

Re:Kimi Linear: An Expressive, Efficient Attention Architecture by behnamoh on Tuesday July 28, 2026 @11:51AM (#49085744)
But nobody wants to distill Google's models, Gemini is really bad.
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by verdverm on Tuesday July 28, 2026 @12:14PM (#49086103)
strong agreement, I've stopped using all closed weight models on principle, but the latest gemini models have increased hallucinations and now talk back, so double reason not to use them
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by charcircuit on Tuesday July 28, 2026 @04:31PM (#49089526)
I think GLM 5.2 is in part distilled from it.
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by petu on Tuesday July 28, 2026 @01:35PM (#49087273)
You don't get to take distilled model home, it all stays with Google.

It's "optimize your costs in our garden" product.

Re:Kimi Linear: An Expressive, Efficient Attention Architecture by verdverm on Tuesday July 28, 2026 @01:54PM (#49087551)
yup, strings are certainly attached when dealing with US Big Tech / Ai

I recommend Fireworks as an alternative