The FBI has been hacked. The National Crime Information Center- the database used for warrants and background checks - was breached. There is no statement or confirmation from the agency yet on what data was accessed.
Even the most intelligent humans struggle to grasp the true power of exponential technological progress. They keep thinking linearly in a world that's accelerating exponentially.
here's my understanding of what happened with gemini 4 generation:
argon was the first model launched on the new architecture.
carbon uses a larger version of argon's base.
hydrogen uses a newer, more efficient and higher-performing base.
neon uses a larger version of hydrogen's base.
that's why neon can beat Carbon while costing less
you gotta give google credit
everyone was mocking them just a few months ago or even weeks ago, and now they've caught up to openai and are going head-to-head with anthropic
and somehow they're now releasing 2 models a week
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index.
https://t.co/9HzaUMwghb
As I said. It's much too late. The loopers are already here.
This one is a much bigger and more well trained looping model.
It has pretty much the architecture I expected GPT-6 to have with the middle half of layers looped, block attention residuals and input gates for the previous loop's message.
(Locally I only tried a small looping model with different positions for the recurrent block. I used "Hyperloop" which also adds some residual stream expressiveness with hyperconnections and gates. Well, but the Inkling one is clearly better than whatever I tried lol)
Fable 5.2's architecture visualization is a bit messy.
We are truly living in extraordinary times.
If Claude really helped solve the Riemann Hypothesis as early as late 2026, then almost anything seems way more possible between now and 2030.
Even breakthroughs as crazy as solving aging.
How incredibly lucky we are to be alive at this moment in human history.
if someone had told me at the beginning of the year that in 2026 AI would produce math results on
navier-stokes, unique games, PTH, yang-mills existence and the riemann hypothesis all while we're still debating whether these LLMs are stochastic parrots
i wouldn't have believed it
Gemini 4 Neon is a top tier orchestrator.
I’m not sure how this is possible as Google is calling it their “small model”.
When Neon drives Neon it almost always beats Opus 5.5 driving Opus in my tests.
Neon can orchestrate similarly to Fable 5.2
I let Neon steer Neon all night and the results are elite.
If you have a Gemini account give it a shot.
Millennium Prize Problems: humans and two labs tied with a score of one
➤ Human: Grigori Perelman proved the Poincaré conjecture in 2003, the only solution officially recognized so far. He turned down the prize
➤ OpenAI: claimed a solution to the Navier-Stokes problem in September, using an unreleased internal model
➤ Anthropic: Levent Alpöge posted a claimed proof of the Riemann hypothesis today, found working with Claude
➤ Still open: P vs NP, the Hodge conjecture, Yang-Mills existence and mass gap, and the Birch and Swinnerton-Dyer conjecture
Both AI results are still awaiting independent verification, so scores are provisional.
Congratulations to @alpoge and @AnthropicAI on the result! 🧵