Was vibe coding over the weekend, just shimmying through some ideas, pushing in on some fluttering, and after some meditation over code samples I had to prestidigitate my way back to the core focus and avoid distractions. Suddenly I found I was Julienning! And now I get Dario!
@joedaroo Blew my mind a bit here and that’s good. I think most of us approach security as a binary model, when all along we should be thinking ‘quantum’. Excellent post!
Monday blues redefined… Claude Opus 5: ‘Thinking’…
An hour later… ‘Almost done thinking’… even on the tiniest task. CoT gone apparently (yes yes Fable but not on my budget)
@BrianCoryDobbs@grok ‘s response: While the LR results were positive, Viking’s other two biology experiments yielded negative or ambiguous outcomes, leading NASA to favor non-biological explanations like soil oxidants, though Levin maintained a biological interpretation for decades.
We love the idea of @bot, but is it aimed at the right market? It is literally the best tool any AI company has released to date to bridge the gap between the masses of AI ignorant people and the very few in Grok Heavy, who all can build everything Bot does anyway @SpaceXAI
@demishassabis Very well written and considered post, Demis. I would add that we now have good examples of growth v non-growth, freedom v control when we consider the US & EU regulatory outcomes, the learnings of which should be a caution to AGI ‘management’.
Oh, but then you do, AFTER the auto reload has been created. So many more of us customers would sign up for auto load if we knew about the budget setting upfront when first activating the auto credit reload 😊
I really think you can do your credit / purchase user journeys better than your current one, @AnthropicAI If you have to spend pages of your wiki to explain how things work, then perhaps a redesign is called for.
However, given how LLMs plough through credits at will, using only the above settings may result in the billing of multiple recharges per month for a user that could completely max out their card. Why not just add a total budget that disables auto when reached?
Somewhat frustrating to be an @AnthropicAI subscriber in the UK + monthly credit purchases, then loose access to the top model, only to still face a usage cap for older versions like Claude Sonnet 4.6. Seriously? 😑
If you work in fintech and don’t understand these terms, you’re building blind.
Here’s A–Z of fintech terms that actually show up in real systems (not textbook definitions, but how they behave in production).
This morning RoW wakes up to the reality that consuming ‘free’ US services for years & using those same ‘free’ services to attack the US relentlessly whilst producing no competitive service, other than volumes of regulations, has consequences, even for 4 of the 5 👁️s’.