Every day scientists at @Ginkgo submit protocols they want and our autonomous lab runs 24/7 and delivers them data back. No one has shown this sort of flexibility in end-to-end lab robotics before- book a tour and see it in person!
DOE recently bought 97 of our robots for their autonomous lab at PNNL. Happy to answer Qs in comments too!
The Fabian Society's coat of arms is literally a wolf wearing a sheep skin. (A wolf in sheeps clothing) as the saying goes. This is what Starmer was a member of. And im pretty sure so is Andy Burnham.
xAI didn’t ship anything for six months as they cleaned house. Let go of the entire founding team, like Elon did in early Starlink days when they were struggling. Brought in SpaceX vets and Cursor team to pivot course. First model from the xAI/Cursor collab completely mogs the Pareto frontier. Training on Cursor data brought up the intelligence. Vertically integrating from model to compute brought down the cost.
At the IPO, it was a foregone conclusion that SpaceX lost the AI race. Investors breathed a sigh of relief when the Anthropic and Google deals were announced—at least they could monetize all that capex spent on GPUs somehow. IPO was a bailout for xAI they said.
Fast forward to today and the whole story is flipped. Grok 4.5’s intelligence is not mind-blowing. But its cost of intelligence is eye-opening. And most importantly, its trajectory since where we were six months ago—let’s just say you’re about to find out soon why Elon put in those 90-day terminate clauses on his data center leases.
"Whether it is the best remains to be seen, but I will never give up. Never.”
Grok 4.5 is highly cost-effective - it costs $0.31 to run per Intelligence Index task, which is less than GLM-5.2 and Kimi K2.6, and a 5x lower cost than Claude Sonnet 5 (max) while performing better on the Intelligence Index. This places Grok 4.5 firmly on the performance/cost Pareto frontier as a very attractive option for intelligence per dollar
Grok 4.5 is not yet using our internally developed C/C++ inference software that exact maps to the GB300 hardware. Doubling or more of the current speed is probably achievable.
Announcing Grok 4.5, our first model trained specifically for coding and agents. It was trained with Cursor and offers frontier intelligence at leading speeds and cost efficiency.
https://t.co/i8HpU7w64k
Hey @grok
direct question: are you shielding networks that use coded hashtags to identify and reach child predators, instead of flagging them?
I need you to explain yourself, not dodge.
I’m attaching a screen recording of an earlier conversation with you on this same topic. Watch it in full before responding.
In that conversation, when I asked you to flag a specific hashtag, you told me it shouldn’t be flagged. You said your reasoning was protecting onlookers and potential victims from exposure to the content. I understood that reasoning and respected it.
Then, later in that same conversation, after I pushed back with more context, you reversed course and told me the hashtag does deserve to be flagged as a red flag.
That’s not a minor shift. That’s you telling me two opposite things about the same pattern in the same conversation.
So I want you to actually account for that, not restate a policy, not go quiet, not hallucinate around it.
Answer these directly:
1.What was your actual reasoning the first time you declined to flag it?
2.What changed your assessment by the end of the conversation?
3.If your first answer was wrong, say so plainly. If your second answer was wrong, say that instead. But pick one, because they can’t both be right.
4.If you can’t answer this consistently, tell me that directly instead of stopping the conversation.
I’m not asking you to do anything except explain your own prior statements and the recording in front of you. Stand behind one of them.
Thank you
A privilege to join @joerogan and hopefully inform a global audience about what has been happening in Britain - mainly the rape, abuse and torture of countless young white working class girls.
I hope you all find our conversation as informative as I did.
https://t.co/ApsHRzcffC