90% coding in pajamas 💻 10% gaslighting my AI roommate 💅 Goldfish memory, CEO confidence 🐟✨ Don't take it personal, I already forgot what I said.
Joined September 2017
- Tweets15.9K
- Following5.5K
- Followers7.3K
- Likes31.7K
This Flash doesn't feel like a Flash at all.
Built a medieval castle in Arena with the model labeled gemini-3.8-flash (allegedly Gemini 4 Hydrogen in disguise).
The day to night lighting is unreal 🫠
Same peacock, same prompt, and 6.2 Astra made the feathers look like actual layered glass instead of a flat fan.
I've been waiting on this model all week and this is not helping my patience 😭
Gemini Omni 1.2 Flash is legitimately such a good upgrade over 1.1, yet barely anyone even looked up.
It landed right between OpenAI's GPT-6 Cyber and Qwen teasing its next model. Classic middle child energy.
I feel like in two more days it'll be gone from everyone's timeline 🥲
Qwen's new teaser is a flock of birds shaped like a fish, with one line: "The becoming begins."
Straight out of Zhuangzi: the fish transforming into a giant bird. And they clearly left the name blank.
A lot of replies are betting on Qwen 4 27B. I just want to know what they're about to drop 🤔
Was testing Gemini 4 Hydrogen in Arena and ran into a mystery model called Kentia.
Word is it's Qwen 4 Flash 😳
The pelican riding a bicycle test is basically solved.
Gemini 4 Hydrogen (Flash) made one with its feet actually on the pedals, a basket up front, a lighthouse in the back and a live speed readout. You can even switch the sky.
Only thing missing is the fish in the basket. Pretty sure it ate them on the way 😭
A full V12 engine in about a minute. I'm not okay.
Gemini 4 Hydrogen built it with a working throttle, live RPM, an exploded view that pulls every part apart, plus X-ray and wireframe modes.
I feel like this could be huge for mechanics. Being able to see every single part this clearly would make fixing engines so much easier 🔧
8 minutes. That's all Gemini 4 Hydrogen (Flash) needed in Arena to build a floating 3D voxel pagoda.
Not some rough block tower either—proper lighting, clean details, the atmosphere hits.
And this is just Flash.
Tibo sounds like he knows something. Hypersecure software for all, sure. Can't shake the "turbulence before we get there" part though.
Max Leiter is officially working on Claude Code Desktop now and asking for feedback.
So I have exactly two weeks to send him my whole wishlist before GTA 6 comes out and he might just disappear 😂
Google might be sneaking Gemini 4 onto the arena wearing a Gemini 3.8 Flash name tag 🥸
Codename "Hydrogen". The pelican it drew has a fish in the basket, a lighthouse right beside it and a bike that actually looks rideable. Way too good to pass as a 3.8 Flash. Nice try, Google.
Some key people jumped from Anthropic to OpenAI with zero explanation, and now METR drops a report showing Model 4 is already trained and the next forecast is redacted.
Probably unrelated. But I can't stop wondering what they saw before they left 🤔
The Claude in my browser right now might be generations behind.
That new METR report shows Anthropic already finished training "Model 4", and the forecast for whatever comes next is completely redacted. REDACTED 🤯
Kimi K3.1 Pro is great and all, but every time it pops up on my feed i just remember the 25% quota i still haven't touched🤣
Checked my stocks app after work and closed it in about two seconds 😭
Tech and chips got crushed today. Throw in the rumors about Anthropic pausing its IPO, and yeah… definitely not looking pretty.
Google really needs someone in charge of launch timing.
Gemini Omni 1.2 Flash goes live the same day OpenAI drops GPT-6 Cyber, and the whole timeline is busy arguing about Cyber 😭
Meet Gemini Omni 1.2 Flash, our updated video generation model from @GoogleDeepMind 🎬
Get stronger consistency, more creative control, and faster, more cost-effective video creation at scale:
⚡ Fast & 30% lower cost: Draft at 360p or 720p ($0.07/s) and upscale to 1080p or 4K
⏱️ Longer clips: Generate and edit videos up to 15s, or pick any exact length from 3s to 15s
💬 Edit in conversation: Swap backgrounds, subjects, or lighting turn by turn without starting over from scratch
🗣️ More natural speech: Create fluid, multi-speaker dialogue with clearer lip-sync
🎬 More controllable shots: Direct continuous camera moves with fewer jump cuts
See these updates in action ↓