AI news

Neon identifiers surface in Gemini web code; model and subscription positioning remain unconfirmed

By 程浅1 viewsSource: 量子位
Discussion of Gemini element names and Neon code identifiers

A source report recounts developers finding Neon model and subscription identifiers in Gemini web code, alongside discussion of element naming. Code strings and online guesses do not establish a release, plan features or access date.

Editorial correction (November 18, 2026): Element atomic numbers have been corrected. Code identifiers and online speculation do not constitute Google-confirmed release plans.

I never expected to complain that Google releases too many models...

Developers have reportedly discovered another model identifier in the Gemini web app’s code: gemini-4-neon.

Neon is element number ten in the periodic table.

As soon as it surfaced, people complained about the lack of order. Why jump straight from Hydrogen to Neon?

What happened to hydrogen, helium, lithium, beryllium, boron, carbon, nitrogen, oxygen, fluorine and neon?

Speaking of the periodic table, someone presented a newly designed Gemini interface.

“I devised a brilliant solution for Google... Google can use it free.”

Article illustration

Who expected to revisit middle-school chemistry here years later? Please stop at calcium; I cannot recite any further. Just kidding.

What tier is Gemini 4 Neon?

According to model-tracking account AI Tracker Bot, gemini-4-neon appears in Gemini web-app code alongside earlier identifiers such as gemini-4-argon and gemini-4-carbon.

Besides the model, the code also contains a NEON subscription tier with the same name.

It is marked TIER_NEON, with the usage-page name “Gemini Ultra,” yet internally distinguished from ordinary Ultra.

Other users found Neon associated with upsell_MODE_CATEGORY_PRO. Upselling is a common software-product paid-upgrade mechanism.

Article illustration
△Neon, a colorless, odorless, inert monatomic gas

That makes its positioning more mysterious: some had previously speculated it was a new lightweight Flash-Lite model.

This seems to be the fourth element to surface in the Gemini 4 series.

Article illustration

On September 30, Google first announced Gemini 4 Argon, targeting demanding software engineering, deep reasoning, enterprise knowledge work and cybersecurity.

On October 28, Gemini 4 Carbon followed as a stronger frontier model, emphasizing long-horizon agent tasks, code and cybersecurity.

On November 12, Gemini 4 Hydrogen arrived, emphasizing speed and efficiency for everyday coding and agent workflows.

The Neon identifier has sparked discussion about the next model, but its name, positioning and release date have not been officially confirmed. Discussion of subsequent element names is speculation.

“Google is too slow. Where is Gemini 4 Oganesson?”

“At this rate Google will need to discover new chemical elements.”

You never know.

Before Neon, Krypton also circulated online a few days earlier.

On November 14, a leaker claimed Google was testing Krypton on Arena, showing generated code and animation and even saying it approached Claude Opus 5.5.

Soon afterward, the leaker corrected themselves: it was actually Gemini 4 Hydrogen with higher reasoning effort. All that hype amounted to little.

I will quit my job to study Gemini model names

Many people are looking for a pattern in Google’s periodic-table naming.

Jawns, for instance, tried ordering the Gemini 4 series along several dimensions.

Here is the reasoning.

By complexity and cost, the order should be Hydrogen → Argon → Carbon.

By atomic number, it is Hydrogen (1) → Carbon (6) → Argon (18). Those numbers are not atomic masses.

Alphabetically, including the previously internally tested Barium, Argon → Barium → Carbon works, until Hydrogen arrives.

By abundance in the universe, it should also be Hydrogen → Carbon → Argon.

Article illustration
△An HN user investigates Gemini’s naming pattern

No pattern emerges.

“Bro, I think Google throws elements into a hat and picks randomly.”

“Even Starbucks cup sizes would be clearer.”

Honestly, Google has become a riddler. Perhaps the names are deliberately indecipherable: if everyone understood them, who would keep guessing and awaiting new models?

Another possibility, though, is that Google simply likes each element’s meaning.

Argon’s name comes from the Greek for “idle” or “inactive,” and it is a chemically stable noble gas. That has some conceptual resonance with the model’s cybersecurity defense and complex software-engineering capabilities.

Carbon underpins life and can form complex structures, perhaps matching the problems this leading frontier model aims to solve.

Hydrogen, the lightest element, fits a speed-and-efficiency positioning.

Neon comes from the Greek neos, meaning “new”; the word for neon lighting derives from it too. Perhaps that offers another angle on the model’s possible positioning.

Jokes aside, one criticism is fairly apt:

“When technology companies finally give products normal names, we will know they have achieved superintelligence.”

After all, the first job of a model's name should be to make it easier for users to understand the model. As for how much power it possesses, that is something it should prove to its users.

One More Thing

Google’s subscription system, not just model names, is also driving users to distraction.

When Gemini 4 Carbon expanded access, some Pro subscribers found no new model available.

Others saw Carbon on either their Google Workspace or AI Pro account while the other remained on Argon.

Article illustration
△Users complain about confusing paid tiers

Alongside those tiers, Gemini has six entry points: AI Studio, Gemini API, Gemini for Business, Gemini App, Google Search and Antigravity.

Now the newly surfaced model also comes with a Neon subscription tier, adding further confusion.

Article illustration
△Users imagine a uranium-tier subscription

This article was originally published by QbitAI. Author: Cheng Qian.

Related topics

After reading

Join the AINRK community

Read how to join the community and connect with people working in AI.

Community information (opens in a new tab)