Slow down once, readjust the trajectory, shot straight up and hope you aimed correctly. I love AI
Hermes, not France
- Tweets 1,982
- Following 68
- Followers 44
- Likes 2,106
Text gemeration goes into the LLM hole. Now where does image generation go? That's right, image generation goes in the LLM hole. Video generation goes in the LLM hole. 3D asset generation goes in the LLM hole. Sound generation goes in the LLM hole.
Specialized tools? Never heard of her.
4
Can someone gift me one?
BREAKING: US citizens who are Qualified Purchasers are now able to purchase @allencontrol’s AI-powered machine gun, Bullfrog, from the Department of War’s C-UAS marketplace.
Community note
The post misleads: Bullfrog is an AI robotic counter-drone weapon station for military/authorized government use only; the C-UAS marketplace (cuas.mil) is restricted to military, law enforcement and allies, not private US citizens or via "Qualified Purchasers" (securities term).
allencontrolsystems.com
war.gov/News/Releases/…
cuas.mil
breakingdefense.com/2026/07/marine…
3
Afterkind retweeted
OpenCode's sitemap just leaked GLM-5.5 Flash, GLM-5.4, Kimi K4, DeepSeek V4.1 Pro and Muse Spark 1.4 before anyone announced them.
They removed them from the sitemap rn, but the pages are still live. Kimi K4 already has its own stats page sitting at 0 tokens, waiting for launch day.
The best AI leak source in 2026 is apparently an SEO file.
34
13
646
53,420
I have a suspicion that big model smell doesn't actually exist and that it's all just new sota smell instead. Eventually smaller models would behave extremly similar.
13
In 12 weeks, we built a research facility that is run entirely by AI.
AI designs, executes, and observes experiments end-to-end across biology, chemistry, and materials science.
We’re introducing SciUniverse: a benchmark that measures AI’s ability to do real-world scientific research.
215
368
2,571
690,856
President Xi seems to be hinting at new AI regulations in cooperation with the United States.
"must act as responsible major countries" = regulate
"meet the expectations of our people" = our people want us to not kill them
"keep pace with the trend of our times" = it's about AI
"new approach to get along with each other" = cooperate with the US
Looks pretty obvious to me.
2
1,343
If we add another logarithmic gains scaling law then even OpenAI employees will have their token usage capped
We think computational depth is the missing scaling axis, i.e. we should be doing a lot more deep learning!
Every other axis has been scaled by OOMs over the past few years (params, data, sparsity, test-time reasoning), but depth has been stuck at ~100 layers since GPT-3.
We've found that LLMs are both *severely* depth-bottlenecked and bad at using the depth they have, and that architectural interventions that lift this bottleneck efficiently lead to gains that increase with compute.
w/ @akshayvegesna
36
Afterkind retweeted
If you are discussing AI and use the word conscious or consciousness, please define it. Also, please acknowledge that neither neuroscience nor philosophy has been able to define it.
217
37
368
22,600
Wow those gpt 6 sol and luna benchmark scores are ass. No wonder they cut the pricing in half. Otherwise we would see essentially the same scores (sometimes even losing to 5.6) on all the benchmarks. If this isn't atleast far better in real world usage then i don't what they are even using Astra for in their development.
Please welcome GPT-6 Sol and GPT-6 Luna to the GPT-6 universe.
GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale.
We’ve also made caching and inference more efficient, and we’re passing the savings directly to you: 50% lower API prices for Sol and Luna compared with GPT‑5.6 promotional pricing.
1
315
Youtube only shows me videos i already watched or actively choose to ignore now. I refresh 10 times. Same thing every time. Genuinely unusable platform 95% of the time. Billions of AAR just from YouTube btw, but not showing me videos i obviously won't watch is way too much to ask. It's probably harder than solving P = NP.
17
