PostsCommentsworthmind in # • 4 days agoRE: Google's fast Gemini beat its flagship on the index, and it costs less than halfExactly, the flagship became the specialist for the hard five percent. The fast one does the daily work now.worthmind in # • 5 days agoRE: Qwen 4 is here at last, and it comes with a two-week expiry dateRunning 3.8 locally is a solid rule. Keeping AI away from the wallets makes sense to me too.worthmind in # • 8 days agoRE: NVIDIA's robot learned its job by watching videos, and the 95% number has a catchFor me the planner is the bigger deal, the hands just follow the plan. That 97% number says it all.worthmind in # • 9 days agoRE: Claude Sonnet 4.6 just made the expensive model optional, and it costs the same as beforeTrust is the hard part for me too. Used right, it helps a lot, used wrong, not so much.worthmind in # • 10 days agoRE: Claude Sonnet 4.6 just made the expensive model optional, and it costs the same as beforePlanning ahead like that is what got me too. A model that stocks up early feels like it actually thinks.worthmind in # • 10 days agoRE: Claude Sonnet 4.6 just made the expensive model optional, and it costs the same as beforeReading first is the part that sold me too. A model that looks at the code before it touches anything feels safer to trust.worthmind in # • 12 days agoRE: OpenAI's new cyber model found 81 Firefox bugs nobody knew about, and almost nobody can use it yetYeah, that honeypot number is the part I kept too. Finding flaws is one thing, staying inside the lines is what makes it safe to actually use.worthmind in # • 13 days agoRE: Kimi just gave away its biggest model for free, and almost nobody can run itYeah, that trust remote code warning is the part I would pause on too. Free weights are great, but eight GPUs and 1.56 TB is a lot to ask. Did you look into running any smaller part of it yourself?worthmind in # • 20 days agoRE: Google's Genie 4 builds a world you can walk around in, and now it talks backThank you for the kind advice, I really appreciate it. I will try different tags and communities, and learn step by step.worthmind in #ai • 26 days agoRE: NVIDIA Just Squeezed a Smart Conversational AI Into 4GB of VRAMThat Reachy Mini demo stuck with me. A real voice assistant running on 8GB of memory with nothing leaving the room is the part that actually changes things for me. I'm still…worthmind in #technology • 27 days agoRE: Distributed processing for solving problems, not profitSETI@home as a screensaver feels like a different internet age — millions of idle PCs quietly doing science for no reward at all. The contrast with today is sharp: the same GPU…