Did you use it in anti gravity? I feel like it's held back alot by the harness
tried out gemini 4 neon today
it does not beat opus 5.5. it is good, but does not understand my intent as well as opus, and is not as thorough.
for example, i asked it to add a diagonal strikethrough in a box in a web app when that box was disabled, and it failed to do it properly the first time (the line did not reach both corners properly). no doubt opus would have nailed this first try and proven it with screenshots.
neon is very eager though, and it's kind of fun. i still need to do more testing on it to see what it's capable of.
1
Exactly as I thought. Antigravity has a lot of improvement that needs to be made @antigravity
1
7
If this is true GLM5.5 it's due to them doing their own Data Collection and cleaning and not doing any distillation at all. Calling it now.
Not sure if it's true.
But if Gestalt is GLM-5.5 then Zhipu has definetly cracked something internally.
It is a massive jump, how big that model has to be?
So many questions, hope it will get answered soon !
174
Witizenship retweeted
We collaborated with @AMD to bring local model inference to AMD NPUs.
NPUs deliver powerful compute performance while using 2–4× less power than traditional GPUs.
We teamed up to enable running models with less power!
Strix, Gorgon, Krackan & more.
Blog: newsroom.amd.com/news/amd-un…
AMD XDNA 2 NPUs are the world's first NPUs supported by @UnslothAI Desktop. 100+ agentic apps. Now on your NPU. bit.ly/4jqlftF
Witizenship retweeted
I have been testing the new Gemini 4 Neon, but the usage is quite low for my Pro account.
If they can fix that, i can see this being used by everyone.
Or let us use it with any harness
Google has one thing left to fix.
Antigravity.
It's weak compared to Claude Code or Codex.
As someone who loves desktop apps, it's not customizable at all and doesn't even feel strong as a harness.
Either merge it into the Gemini app or lock in for the next few weeks and give us a new Antigravity experience.
22
Why is my Argon usage going down when I am using Hydrogen? @GeminiApp
10
I said it before, AMD should be training models specifically on all of their documentation and IP, and using that to optimize things.
Working with @AMD, our models identified up to 14% PPA improvement on flagship GPU IP in less than five months—building on designs AMD’s engineers had already spent years optimizing.
The same approach helped improve Jalapeño, our first custom inference chip, with gains verified on real silicon.
New evaluations also show our models getting better at certain chip design tasks.
We’re sharing these results and how we’re working with @Synopsys to bring our model capabilities and their trusted EDA design and validation tools to more chip designers.
openai.com/index/advancing-c…










































