Google's Gemini 4 Carbon tops three agent benchmarks at less than half the cost of its closest rival
Gemini 4 Carbon is now available
Google has opened Gemini 4 Carbon to developers, enterprises, scientists, and consumers. The model ranks first on CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench while costing more than 2X less per task than the next best external model, and it ships at Argon's introductory price of $2 per million input tokens and $10 per million output tokens. Carbon also delivers Google's strongest cybersecurity results to date, hitting 100% recall on scanning in Project Fortify.
When scanning two major SaaS providers, the model uncovered 0-day vulnerabilities in under two hours, compressing weeks of vulnerability research to a single afternoon.
- byefruit
Just checked my Gemini app, still only have 3.1 Pro, 3.6 Flash and 3.6 Thinking on Pro. Looks like it's Ultra only.
They've been so slow to update their top models and roll them out that despite subscribing to the Gemini Pro plan, I use it so rarely that this is my last month. That and their product offering was so confusing I basically gave up.
- 5555watch
Available through the API.
To add, it's actually surprisingly good. I have this unfinished paper draft that I keep feeding to various top AI's to see what it picks up on and what suggests to edit, and while Astra/Fable/Opus/Argon so far has been mediocre, Carbon's answer surprised me with some interesting points all the other AI's missed.
It's an anecdote, sample of one, but it got my attention.
- simonw
Pelicans are solid, but every Gemini model for the past year has produced solid pelicans: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...
I also had it do this in Antigravity:
> Make a three js animation where yonkers stands up as a big mech (cool stand up motion) and stomps around manhattan and queens and Brooklyn, use openstreetmap mapping data
The result was pretty great: https://static.simonwillison.net/static/2026/yonkers-mech-st... (I had to prompt "build that into a standalone HTML file" to get that version").
Here's the same prompt (except I had a typo, I put "átomos" instead of "stomps", it figured out what I meant) against Fable 5.2: https://tools.simonwillison.net/yonkers-rises
- guidopallemans
A nice nod to simonw in one of the highlight videos;
- dom96
It is such a massive miss from Google that it took them this long to release this model. I just ran it on my benchmark[1] and it matches Fable 5.2. They could have been first to get there, why didn't they?
That being said, it is still slower and more costly than Fable 5.2. So Anthropic wins in my eyes.
- markush_
I have google AI plus because of drive storage and it’s ridiculous that the best model this gives access to is 3.6 flash, not even the latest flash model. Google either needs to stop this being branded anything “AI” or make better models available if this is what the entrance subscription gives
- djoldman
> Getting started with Gemini 4 Carbon
> For developers: Build with the Gemini API paid tier in AI Studio and Android Studio, and if you’re a Google AI Ultra subscriber, you can try it in Google Antigravity and AI Studio.
> For enterprises: Try it in Gemini for business via the API or model drop down, Google Antigravity, or CodeMender.
> For everyday users: Try Carbon starting today in the Gemini app and AI Mode in Search if you’re a Google AI Ultra subscriber.
> For scientists & researchers: Apply for a chance to get access to both Argon and Carbon via Google Antigravity’s highest usage tier free of charge for 6 months (onboarded on a rolling basis), subject to eligibility, capacity, and selection.
> For cybersecurity defenders: Apply to join the Fairwind Program to get access to Carbon for cyber on a rolling basis.
Looks like the only way to access any Carbon tokens is to pay $99 / month, pay via API, or apply as a scientist.
- nothercastle
Is it 100$ a month for the minimum plan that has access