Pinned Tweet
This is a big checkpoint for me, I created this account 2.5 weeks ago and now I have 1M impressions and 650+ followers without any viral or engagement bait posts, pure organic growth.
Thanks to everyone who followed and engaged with my posts. Onward! This is just the beginning.
camper π¦π£ retweeted
Chinese AI labs: Chill, we still have Google Neon.
Distillation can reproduce powerful AI capabilities without the safeguards built around them.
Weβre sharing new research on our work to combat distillation of our models, how these capabilities transfer, and how we detect and disrupt coordinated attempts.
These findings are helping us sharpen our defenses and support safeguards across the industry.
openai.com/index/combatting-β¦
camper π¦π£ retweeted
π I asked GPT 6.2 Astra Xhigh to put me in a wooden boat in a storm. The first ocean looked cartoonish. Six visual passes later, the last wave hides the horizon.
β±οΈ 47 minutes of build time, including renders
πΈ $14.85 at API rates across 159 model calls
Sound on for the thunder. #ThreeJS
camper π¦π£ retweeted
Big update for Plus users and a visual overhaul that looks great actually!
NEW:
- ChatGPT Space for Plus!
- new, simpler visual design across the app
- huge performance improvements everywhere
- completely rebuilt threadsβ better perf and design
- redesigned Space home
- much faster pages and slides
- edit pages on mobile
- more reliable Codex Remote
much more to come soon!
camper π¦π£ retweeted
This is the best suggestion. Please make the usage unified and then it'll increase in general. I think many people would love that.
Suggestion for @GoogleDeepMind that many people would love to happen:
Unify the usage limits across your services
Antigravity, AI Studio, Gemini App
There's a lot of usage but scattered across all of them and many of us would love to allocate all of it in what we use the most.
Whether I wanna burn all of it in the app, studio or antigravity I should be able to.
I swear OpenAI and Anthropic internally communicate when to release their things so they can co-ordinate lmao.
Excited to share that a new, refreshed mobile experience for the Claude app has started rolling out!
We wanted a design that better supported code, projects, and artifacts on mobile, and made it easier than ever to start and resume sessions on the go.
With this refresh, we also took the time to refine so many parts of the app and introduce quality-of-life features.
This is a continuation of all the great work the team has been doing on desktop, and weβre not finished yet!
Huge thanks to the team @pranathiperii @khalilcader @RobertJBye Andrew, and Graham for all their work on this, check out their posts for more details!
3
5
506
Chinese AI labs: Chill, we still have Google Neon.
Distillation can reproduce powerful AI capabilities without the safeguards built around them.
Weβre sharing new research on our work to combat distillation of our models, how these capabilities transfer, and how we detect and disrupt coordinated attempts.
These findings are helping us sharpen our defenses and support safeguards across the industry.
openai.com/index/combatting-β¦
π I asked GPT 6.2 Astra Xhigh to put me in a wooden boat in a storm. The first ocean looked cartoonish. Six visual passes later, the last wave hides the horizon.
β±οΈ 47 minutes of build time, including renders
πΈ $14.85 at API rates across 159 model calls
Sound on for the thunder. #ThreeJS
camper π¦π£ retweeted
βοΈ I asked Gemini 4 Neon to drop a meteorite from space to Earth in pure Three.js.
Zero 3D assets, zero external physics sims β every shader, rock & cloud is code, paired with 29 real ASMR recordings. π€―
β±οΈ 55 minutes (incl. 1080p60 GPU render)
πΈ $24.67 in API costs
π€ Gemini 4 Neon (High) in Antigravity
What it built on its own π
π Ray-traced Earth from 250 km low orbit down to a golden-hour desert canyon with Rayleigh/Mie limb scattering and a floating origin so 6 orders of magnitude stay jitter-free
πͺ¨ Sculpted an iron-silicate meteorite with regmaglypts, hairline glowing thermal fractures, 13 detaching 3D spalling chunks, and 3,200 velocity-stretched ablation sparks
π₯ Raymarched a 3D volumetric hypersonic bow shock and fiery plasma wake in GLSL, diving straight through 460 backlit golden cumulus clouds into a sweeping aerial pullback
π₯ Built the desert impact purely in shaders: real-time 155 m molten crater excavation, pearlescent Mach condensation dome, ground dust shockwave, 64-step volumetric ash plume, and 900 ballistic ejecta boulders
π Layered 29 real 48/96/192 kHz recordings (100 kHz cracking ice, Space Shuttle launch, Kilauea lava, skydiver chest wind, Mount Yasur volcano, close thunder & falling stones) synced to the timeline clock (-15.48 LUFS, -2.02 dBTP)
π Caught and fixed flat non-indexed asteroid facets, plasma white-out, and a signed-field plume bug during its own look passes before the final render
π¦ 20 s Β· 1920Γ1080 Β· 60 fps (SS=2) Β· 29 real recordings Β· 0 external assets Β· 23.0 MB MP4
camper π¦π£ retweeted
Everyone seeing news about Reimann Hypothesis but what IS that hypothesis? π
The Riemann Hypothesis
It starts with prime numbers
A **prime number** can only be divided evenly by 1 and itself:
> 2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41β¦
Every whole number is built by multiplying primes together. For example, 12 = 2 Γ 2 Γ 3. That makes primes the building blocks of numbers, a bit like atoms are for matter.
The puzzle: primes look random
Here's how the primes are spaced:
- 2, 3: right next to each other
- 23 to 29: a gap of 6
- 113 to 127: a gap of 14
You can't predict exactly where the next prime will be. But when you zoom out, a pattern shows up. Primes slowly get rarer, at a very predictable *average* rate. Near 1,000, about 1 in 7 numbers is prime. Near 1,000,000, it's about 1 in 14.
So primes look messy up close and orderly from far away. The real question is **how far the actual primes stray from that average.**
Riemann's discovery (1859)
Bernhard Riemann found a special formula called the **zeta function**. Think of it as a machine: you feed it a number and it gives you a number back.
He showed that this machine and the primes are deeply connected. The inputs that make the machine output **exactly zero** (called its "zeros") control how the primes are spread out.
The hypothesis itself
Riemann noticed that every important zero he could find sat **on one single straight line**. He guessed that *all* of them do, infinitely many, with no exceptions.
**That guess is the Riemann Hypothesis.**
Why it matters
If it's true, the primes are as well-behaved as they could possibly be. Their wobbles around the average stay small and controlled, like an orchestra where no single note drowns out the others.
If it's false, somewhere far out the primes would wander off their expected pattern in a way nobody expects.
Where it stands
- **Unproven.** It's been open for over 165 years.
- Computers have checked **trillions of zeros**, and every one is on the line. But checking examples isn't proof, because infinity is bigger than trillions.
- It's one of the seven **Millennium Prize Problems**. Proving it earns **$1,000,000**.
- Thousands of other math results begin with "if the Riemann Hypothesis is trueβ¦". A proof would confirm all of them at once.
camper π¦π£ retweeted
Fable 5.5 did pretty good editing!
camper π¦π£ retweeted
Just a reminder, everyone is excited about using Neon in their Antigravity subs. Don't forget, you also get Opus 5.5 and Sonnet 5.5 in that, which is a whole separate limit, doesn't take from your Google models pool!
Big update for Plus users and a visual overhaul that looks great actually!
NEW:
- ChatGPT Space for Plus!
- new, simpler visual design across the app
- huge performance improvements everywhere
- completely rebuilt threadsβ better perf and design
- redesigned Space home
- much faster pages and slides
- edit pages on mobile
- more reliable Codex Remote
much more to come soon!
camper π¦π£ retweeted
Today and NOT by @OpenAI
This is the best suggestion. Please make the usage unified and then it'll increase in general. I think many people would love that.
Suggestion for @GoogleDeepMind that many people would love to happen:
Unify the usage limits across your services
Antigravity, AI Studio, Gemini App
There's a lot of usage but scattered across all of them and many of us would love to allocate all of it in what we use the most.
Whether I wanna burn all of it in the app, studio or antigravity I should be able to.
βοΈ I asked Gemini 4 Neon to drop a meteorite from space to Earth in pure Three.js.
Zero 3D assets, zero external physics sims β every shader, rock & cloud is code, paired with 29 real ASMR recordings. π€―
β±οΈ 55 minutes (incl. 1080p60 GPU render)
πΈ $24.67 in API costs
π€ Gemini 4 Neon (High) in Antigravity
What it built on its own π
π Ray-traced Earth from 250 km low orbit down to a golden-hour desert canyon with Rayleigh/Mie limb scattering and a floating origin so 6 orders of magnitude stay jitter-free
πͺ¨ Sculpted an iron-silicate meteorite with regmaglypts, hairline glowing thermal fractures, 13 detaching 3D spalling chunks, and 3,200 velocity-stretched ablation sparks
π₯ Raymarched a 3D volumetric hypersonic bow shock and fiery plasma wake in GLSL, diving straight through 460 backlit golden cumulus clouds into a sweeping aerial pullback
π₯ Built the desert impact purely in shaders: real-time 155 m molten crater excavation, pearlescent Mach condensation dome, ground dust shockwave, 64-step volumetric ash plume, and 900 ballistic ejecta boulders
π Layered 29 real 48/96/192 kHz recordings (100 kHz cracking ice, Space Shuttle launch, Kilauea lava, skydiver chest wind, Mount Yasur volcano, close thunder & falling stones) synced to the timeline clock (-15.48 LUFS, -2.02 dBTP)
π Caught and fixed flat non-indexed asteroid facets, plasma white-out, and a signed-field plume bug during its own look passes before the final render
π¦ 20 s Β· 1920Γ1080 Β· 60 fps (SS=2) Β· 29 real recordings Β· 0 external assets Β· 23.0 MB MP4
camper π¦π£ retweeted
If Tibo leaves OAI, hope he doesn't join Anthropic or he'll ruin the reputation of quietly destructive launches with his cringe vagueposting.
After Anthropic's IPO, several employees left OpenAI... Something VERY STRANGE must be going on internally.
These are people from safety, robotics, Codex, and several other areas. Is OpenAI doomed to stay in third place forever now? I hope they get through this.
Apparently, even Tibo is tempted to leave.











