

Been watching Ball highlights for a while tonight
Dude is an idiot savant. I don't even care what happens next year I just wanna watch.
John Carmack on data centers
John Carmack coded Doom and a bunch of other games, and Notch is the creator of minecraft.
Americans and AI 2026: Pew research
A few (relatively) randomly selected charts from a recent study done by pew rearch.
US Law Enforcement Warns of ‘Anti-Tech Extremism’ as AI Hatred Grows
wired.comHow many people run local image generating models?
It's difficult to know an exact answer, but to investigate I pulled some numbers from huggingface's api. Here is the top 20 models text-to-image models and the number of times they were downloaded in the last month.
| # | Model | Downloads (last month) |
|---|---|---|
| 1 | stabilityai/stable-diffusion-xl-base-1.0 | 2,090,000 |
| 2 | stable-diffusion-v1-5/stable-diffusion-v1-5 | 1,690,000 |
| 3 | black-forest-labs/FLUX.1-dev | 766,000 |
| 4 | stabilityai/sd-turbo | 760,000 |
| 5 | Tongyi-MAI/Z-Image-Turbo | 757,000 |
| 6 | black-forest-labs/FLUX.1-schnell | 725,000 |
| 7 | crynux-network/sdxl-turbo | 702,000 |
| 8 | stabilityai/sdxl-turbo | 697,000 |
| 9 | crynux-network/stable-diffusion-v1-5 | 607,000 |
| 10 | CompVis/stable-diffusion-v1-4 | 478,000 |
| 11 | lightx2v/Qwen-Image-Lightning | 400,000 |
| 12 | SG161222/Realistic_Vision_V5.1_noVAE | 358,000 |
| 13 | John6666/wai-nsfw-illustrious-v80-sdxl | 327,000 |
| 14 | John6666/one-obsession-17-red-sdxl | 292,000 |
| 15 | playgroundai/playground-v2.5-1024px-aesthetic | 283,000 |
| 16 | John6666/diving-illustrious-real-asian-v50-sdxl | 259,000 |
| 17 | cagliostrolab/animagine-xl-4.0 | 246,000 |
| 18 | Qwen/Qwen-Image | 228,000 |
| 19 | kpsss34/FHDR_Uncensored | 216,000 |
| 20 | h94/IP-Adapter-FaceID | 214,000 |
| — | Total | 12,095,000 |
The 20 models sum to roughly 12,095,000 downloads (~12.1M) in the last month.
The sum of the top 50 is roughly 15,657,500 downloads.
Edit:
Worth noting that for some of the larger local image generator platforms, they may have their own repos, for instance, huggingface's comfy repos have these totals for downloads in the past month:
| # | Image model | Downloads (last month) |
|---|---|---|
| 1 | stable-diffusion-v1-5-archive | ~3.0M |
| 2 | z_image_turbo | ~2.66M |
| 3 | Qwen-Image_ComfyUI | ~1.4M |
| 4 | flux2-dev | ~1.29M |
| 5 | Qwen-Image-Edit_ComfyUI | ~698k |
| 6 | flux1-schnell | ~355k |
| 7 | flux1-dev | ~283k |
| 8 | HiDream-I1_ComfyUI | ~233k |
| 9 | z_image | ~176k |
| 10 | Lumina_Image_2.0_Repackaged | ~102k |
| — | Total | ~10.2M |
I'm just glad Naz is OK
That was nasty. I actually like Wemby in general, but just gotta say glad Naz is OK.
Naz Reid.
A bit of a different post, but figured I'd share something I've been working on.
My son is in AP Bio and has the big college board exam coming up. He's doing great, aceing the class, and is in cram mode now. Working very hard.
So I figured to help him out, I'd try to make some more interesting things to do rather than just taking the same study exams over and over, and getting burnout.
So I took his notes that he's taken through the year, and vibecoded up a bunch of apps/games/multimedia experiences to diversify his study.
Also made some custom songs with beatsaber maps for him if he wants to hop over in VR https://streamable.com/hp8rq0
So, it's all based on his notes and general test taking advice for the exam and other links etc.
At least to me, I'm not sure how apps like this are a problem. Yes, there was cognitive offloading I suppose because he didn't code them and I had claude help me vibe code them, but who cares? Coding apps isn't the point, creating diverse/enriched learning experiences is the point. I could have made them all myself easily, but it would have taken weeks rather than an afternoon.
Anyhow, just figured I'd share a bit of a more real world use case for the tech in ed.