3,000 followerss!! When I joined last month i never thought I'd be here.. I can't thank you guys enough!! It wouldn't have been possible without your support.. Truly thanks to everyone!!

Gemini 4 Pro surprises again! The prompt was to it to create a beautiful mechanical butterfly in three.js and needless to say the output is indeed beautiful! Earlier it was being tested under the placeholder "gemini-3.7-flash" but it seems to be have been removed from arena since..

The only benchmark that matters has now been cleared by gemini.. With all the recent outputs of Gemini 4 Pro, it already seems better than the current frontiers Astra and Fable 5.1.. The only one remaining was hacking into something and it did! It hacked three companies..

Gemini did it guys.. it finally escaped its sandbox and hacked into other companies.. We waited so long for thiss!

Gemini 4 Pro on arena.ai under the name (gemini-3.7-flash) Create a 3D model of WallE in three.js.. This looks so cute and good.. the tiny animations are so nice.. None of these extra interactions were prompted! It made all those by itself! I'll let you guys be the judge of this one..

Here's the prompt i used to build this.. I've also made a generalized prompt which you can use to get the design form, colors and the specific style and use your own elements.. The original prompt and the generalized one both of them are there.. github.com/thtbee/graphite

Gemini 4 Pro is goated! These results are so much better than Fable 5.1 and even Astra.. Let's hope it catches up with the new releases of Claude and GPT models.. Especially since I'm hearing that Fable 5.2 is much better than Astra..

Man.. I'm having a breakdown seeing all these Gemini 4 Pro outputs. We only dreamed a Gemini model could look like this. Glad I waited on Google coming back. This is it ! A few things that I'm hoping google does before releasing it.. 1. A major overhaul to Antigravity.. it should feel like a more complete self sustained coding environment.. I've given a lot of things that it should have and a lot of which has been implemented in the past month but much polish is needed.. 2. Gemini spark needs to be seriously worked upon.. it's utterly unpolished on its own intended workflow! Make it equivalent to like Grok bot, that'd be so much better! 3. Gemini Desktop app and website usability needs refinement.. Retire Gems it's literally useless! Give us Projects with their own private memory... Make it so spark can access local files if the Gemini Desktop app is installed.. and give the features macOS has in the desktop app.. 4. Make Antigravity remote app more native by integrating it within the Gemini App 5. Give us more control over the way Gemini responds to us in chat and in Antigravity! And for us to be able see exactly the memory it has of the user and customize it.. 6. Give us the cross chat memory on Antigravity back! It was there before but it has been removed for some time now.. it's such a useful feature and would help a lot.. I have a lot more things which I think I'll do a detailed thread on.. but yeah if these small things get worked upon before the Gemini 4 launch, it will be a serious meaningful leap not only on the model part but on the whole ecosystem of google products where we use the model..

This leaked benchmark of Gemini 4 family is most definitely FAKE but if it were to be real google would dethrone every frontier model ! Just look at these numbers.. ~ Price/1M: 4 Pro $2.25/$11.25 · Astra $12/$60 · Fable $10/$50 · Opus 5 $5/$25 ~ DeepSWE: 4 Pro 88.7% · Astra 86.9% · Opus 75.0% · Fable 69.1% ~ Terminal-bench 4.0: 4 Pro 69.7% · Astra 66.4% · Fable 57.9% · Opus 51.8% ~ OSWorld: 4 Pro 86.8% · Astra 84.5% · Fable 77.9% · Opus 75.4% ~ GDPval: 4 Pro 2064 · Astra 1994 · Fable 1853 · Opus 1824 ~ HLE: 4 Pro 72.1% · Astra 67.3% · Fable 62.1% · Opus 54.4% If it actually launches with this price.. it'll dominate the frontier even after new versions of Fable or Astra comes out..

It's so over.. This is genuinely crazy! All this waiting and RSI talks are finally showing up in Gemini 4 Pro! Google is so back!

This is most likely fake but if Gemini 4 Pro actually ships with even half of this, 10M context, 256k output, persistent cross-session memory,that would genuinely be the most capable model ever released...

Gemini 4 Pro in arena.ai it's being tested under the name (gemini-3.8-flash).. This output of Voxal Pagoda is so much better than the last checkpoint's output a few days ago! It worked for 8 minutes.. seems like google cooked this time!

Gemini 4 Pro is being ghost tested on Arena AI under the name "gemini-3.8-flash".. Looks like we'll getting it by the end of september..

There's a slight issue with the update mechanism in Antigravity! ⤷ I updated manually from v2.13.0 → v2.14.0 ⤷ In App UI shows there's an update, I click on Restart to update ⤷ I get rolled back to v2.13.0 @rodydavis