Google is internally testing a Gemini 4 Carbon variant that one employee rated close to Opus 5.5 on programming tasks, according to Business Insider. The feedback came from Google’s internal coding platform, where Carbon was deployed over the past few days, and the comparison still needs more testing.
What is Google testing inside Gemini 4?
Business Insider reported, citing documents, screenshots, and internal chats, that Google is testing several Gemini 4 variants named Argon, Barium, and Carbon. Carbon was deployed on Jetski, Google’s internal coding platform, and is said to beat Argon mainly on programming tasks. One employee compared Carbon to Anthropic’s strongest coding model, Opus 5.5.
Earlier Argon builds drew a different reaction. One employee said those versions reminded them of the older Opus 5 on some coding tasks, though internal feedback on Argon was positive overall.
All three names appear to describe checkpoints or updates within one model family rather than separate product tiers. One employee internally called Carbon the "Gemini pro next model," which suggests it would ship under the Argon name. Internal documents also show the now-unveiled Argon was previously called "Barium-B." Whether Carbon ships as an Argon update or a standalone model is still unclear, and there is no announced Gemini 4 release date.
Where does Argon sit in Google’s lineup?
Early rumors pegged Argon as a Flash-style model built for speed over peak performance. Google’s announcement of its Gemini agent for Google Workspace set that aside. The company maps its models to specific roles:
- Argon for frontier reasoning
- Flash for speed and volume
- Omni for generative media
- Gemma for lightweight, open-weight edge workloads
That puts Argon as Google’s most capable reasoning model, comparable to Anthropic’s Opus or OpenAI’s Astra. When Google announced Gemini 4 Argon, it said the model scored 77.9% on DeepSWE v1.1, a benchmark that, in Google’s words, "measures a model’s performance in real-world long-horizon software engineering tasks." Google also said its own engineers have used Argon for daily work, from debugging to large-scale codebase migrations. Carbon and Barium read as iterations on that Argon family.
Why is the update pace speeding up?
The quick update pace, similar to what Google has done with its Flash models, suggests Google has gotten better at using AI to build new models. Google DeepMind employee Vedant Misra responded to the Business Insider report on X by writing, "Have you heard of recursive self improvement," a reference to using AI to help train and refine later models. OpenAI and Anthropic have reported similar progress in their own pipelines.
How is Google preparing its apps?
Without a launch date, Google is already updating the surfaces that would host the new models. The Gemini app now shows an "Automatic" mode for the 3-series models, plus adjustable reasoning intensity from low to high. Google AI Studio has surfaced a new "Ultra" mode promising "advanced skills and tools." Google’s Logan Kilpatrick confirmed the team is working on getting the most out of Argon.
What it means for search and AI visibility
It can be worth checking how easily your content can be found and used online. Google says Argon can "sustain long, multi-step tasks," and it expanded the model’s output token limit to 1M tokens, up from 64K. It can also be worth knowing whether an AI agent can reach and parse your content. Whether an AI agent can read and use a site is what an AI agent readiness check measures, and SEOScanPro‘s AI visibility tracking does this.
What to do now
- Review how your key pages are built and served.
- Review your robots and access rules.
- Track which AI engines currently cite your content, and note the baseline before the next model generation ships.
- Re-run that check from time to time.
- Keep the details a reader looks for in your page text.
FAQ
What is Google’s Gemini 4 Carbon model?
Carbon is an internal codename for a Gemini 4 variant Google tested on its Jetski coding platform. It appears to be a checkpoint or update within the Argon model family rather than a separate product tier.
How does Carbon compare to Anthropic’s Opus 5.5?
According to internal feedback reported by Business Insider, one employee rated Carbon’s coding performance close to Opus 5.5, Anthropic’s strongest coding model. The comparison still needs more testing.
When will Gemini 4 launch?
Google has not announced a launch date. It has already added an "Automatic" mode and low-to-high reasoning controls to the Gemini app and an "Ultra" mode in AI Studio, and the team confirmed it is working to get the most out of Argon.
Try the AI visibility report
SEOScanPro runs a full technical audit of a site and shows the measured result behind every check. Open the AI visibility report.