

Fun fact, two things can be true at the same time. Like corporations can be bad AND people can be stupid.
Mastodon: @greg@clar.ke


Fun fact, two things can be true at the same time. Like corporations can be bad AND people can be stupid.


Our power consumption and the 0bits into USA metric
That’s amazing! Keep us in the loop about the results


I have a RTX4090 (24GB VRAM) plus 128GB DDR4 system RAM on Ubuntu 24.04 and I am running these models locally
| Model | Speed | Context window |
|---|---|---|
| Qwen3.8-Flash-Next UD-Q4_K_XL — 125B/6B-active hybrid MoE, 104 GB | ~16.7 tok/s | 128k tokens |
| GLM-5.3-Flash UD-IQ3_XXS — 320B/18B-active MoE, 120 GB | ~9 tok/s | 128k tokens |
| Qwen3.8-27B UD-Q4_K_M — dense hybrid, 16 GB, fully on the 4090 | ~49 tok/s | 262k tokens |
The larger models don’t fit on the GPU alone but they’re mixture-of-experts and their active weights easily fit on the 24GB. The bottleneck is my system RAM as llama.cpp has to constantly move the experts between RAM and VRAM. I’m running older hardware, AM4 CPU, DDR4 RAM, etc. so 128GB is my limit on system memory and it’s relatively slow compared to DDR5. With DDR5 I would expect a good bump in speed for Qwen 3.8 Flash Next and GLM 5.3 Flash.
I also have a Kimi subscription which I use as my daily agentic driver and use DeepSeek for random tasks (DeepSeek Flash 4.1 is really fast and cheap).


The Anthropic and OpenAI valuations require their models to replace humans. But LLM based AI can’t replace humans because LLMs can’t make decisions. I only use Chinese models now, they’re fast, and they can finish a coding task quicker than I can make an informed decision about the next task.


I’d invest in AI. That’ll get rid of a trillion dollars per quickly


I’m really impressed!


You under estimate my ability to buy stupid things


Oh, border collie! I thought “Big Cock” and was confused. Why would you acronyze “border collie” when there isn’t even any context that you’re talking about a dog?


You can be clear AND concise.


I think I could become a billionaire if I started off as a trillionaire


My argument is that regulation should be designed to prevent immoral behaviour. Otherwise you’re fundamentally taxing morality.


I disagree, divestment in profitable companies just rewards immoral investors. Plus Zuck has full voting control over Meta so shareholder’s votes mean nothing. I just want Zuck to be held accountable under the existing laws.


So basically no consequences for Zuck abusing a generation of children for profit?


I like the music but you’re right, I probably won’t watch 50Tb of talkback tv


I’ll store PBS’s content on my Plex server if they want. I’m sure I’ve got 50Tb free


Ads or promotions may also appear during live events streamed on YouTube, such as sporting events.
I don’t have strong opinions but it looks like they’re referring to this phrasing in the ToS which suggests ads will only appear in sporting events. That’s the ToS connection. But otherwise, they advertise YouTube as an ad free experience which just isn’t true.


You can’t advertise one thing and then just add “nah j/k” in the ToS and think you’re off the hook


That’s driving distance. And if you’re ignoring latitude and focusing on the lower 48 states, why not measure diagonally?


I’m basing this on OP’s comment
Someone in Cuba or Iran can keep installing APKs with no new restrictions, but devs will suffer.
I wouldn’t trust. I’m very lonely