Gemini 4 Argon leads 13 of the 19 benchmarks Google published.
GPT-6 Astra still wins some.
Claude Opus 5.5 wins others.
At this point, asking "Which AI model is the best?" is probably the wrong question.
The better question is:
"Best at doing WHAT?"
Gemini 4 Argon's introductory API pricing is $2 input / $10 output per million tokens.
After November 30?
$4 input / $20 output.
Double.
A reminder for anyone building AI products:
The cost of your AI feature isn't necessarily the cost you'll be paying in production.
Google reports something interesting about Gemini 4 Argon.
35% more agentic code submissions.
30% fewer post-submit rollbacks.
Both observed in its internal Antigravity rollout.
That's more interesting to me than another benchmark victory.
Show me what AI changes in real workflows.
Gemini 4 Argon leads 13 of 19 benchmarks in Google's comparison.
But one number caught my attention: 19.6%.
That's its score on a legal-agent benchmark.
It's well ahead of its rivals, yet still just 19.6%.
Winning a benchmark โ being ready for real users.
Model announcement!!!!!
Google just released "Gemini 4 Argon"
Gemini 4 Argon is now globally available today for developers and enterprises via the API in AI Studio and Gemini platform, for Google AI Ultra subscribers in Google Antigravity, and Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search.