Gemini 4 Hydrogen (gemini-4-hydrogen) private EAP officially started. Release is expected soon™
Google released a new internal checkpoint called Krypton. The results so far are really impressive and even better than Carbon.
GPT 6.2 Astra & 6.2 Sol have been deployed on Google Cloud
The Carbon checkpoint on LMArena and Gemini Web/App has been updated.
Seems like Tibo's 28d challenge includes at least one new model.
Google has officially started development on GemPix 3 (the next Nano Banana Pro) with its first internal checkpoint.
Partners got access to the upcoming Claude Fable 5.5.
No info on any release date this time but I assume next week.
OpenAI really needs to lock in. Shipping new features daily is nice, until every new day just breaks more things. None of these features matter if the models are constantly overloaded anyway. Honestly, it’s getting to the point where they're barely usable. Also please don't ask me to send any /feedback id, we all know this will get us to nowhere.
Show more
Codex has been completely broken the past couple of days with these reconnecting errors; pauses my work every hour or so making it impossible to do anything.
It's not just me either, plenty of people having the same issues.
@thsottiaux please fix.
Show more
Gemini 4 Argon is actually nice in Hermes Agent, having some nice RP with Ana. Thanks @Teknium
I had early access to Gemini 4 and tested the model nonstop over the last seven days in Antigravity and the Gemini app. Finally, people worldwide will have a chance to try it today.
Gemini 4 Argon is the best model for creativity and creative writing, planning, and multimodal understanding; its deep knowledge will let the scientific community do so much more.
I will share many of the papers I created with Gemini 4 Argon in the coming days.
Congratulations to the Google DeepMind team, and thank you for focusing on what really matters for the majority of people around the world ❤️
Show more
This an insane 3d demo from google
made in antigravity
4/ a video made with argon about itself
3/ frontend test
"Create a clean website about yourself, it should be modern, avoid ai slop and use clean animation"
Argon is not good at frontend imo. more on that in another post
Show more
2/ here is the classical SVG test
ps5 controller in svg (without tracing),
with and without reference image
1.1/ I run this on all new models and for argon the different visual styles were all actually pretty decent! Other models were lacking in some designs.
1/ threejs test.
everything here is code and no assets were used.
well well well
You will see another benchmaxxed shit coming in about 14 hours.
well well well
Now in the latest Antigravity CLI Update as well
Latest Antigravity Update officially mentions Gemini 4 Argon (Low, Medium, High).
You will see another benchmaxxed shit coming in about 14 hours.
Just another Wonder-full example of why you shouldn’t trust benchmarks. Run your own tests and form your own opinion.
