AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Why Google’s Gemini 4 Neon Table Needs More Context on ThorstenMeyerAI.com

TL;DR

Google released Gemini 4 Neon on November 18, presenting it as a balance of capability and efficiency. Artificial Analysis scores Neon just one point above Gemini 4 Carbon at the compared settings, while reporting lower task costs and shorter completion times; Google’s benchmark table does not include two models that rank ahead of Neon in that independent index.

Google released Gemini 4 Neon on November 18, describing it as a model built to balance depth and efficiency. In figures published by Artificial Analysis, Neon scores 61 on the Intelligence Index at its xhigh setting, one point above Gemini 4 Carbon at high, while the same comparison shows a lower cost per task and less time to complete it, as discussed in the context behind the AI model rankings. Google’s own benchmark table, however, leaves out two models that rank ahead of Neon in the independent index.

Artificial Analysis reports that a Neon xhigh index task costs $6.44 and takes 21.5 minutes. The comparison Carbon-high task costs $7.76 and takes 37.1 minutes, using Carbon’s standard list price. At Neon’s default high setting, it scores 60—the same score as Carbon in the cited comparison—at $3.74 per task and 14.8 minutes. Artificial Analysis summarizes the release as “a small Intelligence gain over Gemini 4 Carbon at a ~17% lower Cost per Task.” Those figures describe the index’s task mix, not every user’s workload.

Neon is priced at $3 per million input tokens and $15 per million output tokens, compared with Carbon’s standard $4 and $20 rates. Carbon is temporarily discounted to $2 and $10 through November 30, according to the source material, so the standard-price comparison does not reflect that offer. Artificial Analysis attributes Neon’s per-task cost and speed results to its lower token prices and output rate of about 110 tokens per second, compared with Carbon’s roughly 55. Neon also uses about 15% more output tokens per task, offsetting some of the price difference.

Google says Neon supports text, image, audio and video inputs, with text output and a one-million-token context window. It offers four effort levels, from low to xhigh, with high as the default. The model is available through the Gemini API in AI Studio and through specified Google business and subscriber products. Google also says a cyber-defense version will be offered to eligible members of its Fairwind Program over the following few days.

At a glance
reportWhen: Released November 18, 2026; figures acc…
The developmentGoogle released Gemini 4 Neon, a model adapted from Gemini 4 Carbon, with independent testing indicating a small score gain alongside lower cost and faster task completion.

Gemini 4 Neon

Four models in seven weeks

Google’s Gemini 4 releases side by side, each at its best tested setting in the independent Artificial Analysis index. The best value in each row is outlined.

ArgonCarbonHydro.Neon ReleasedSep 30Oct 28Nov 12Nov 18 Index50605461 Cost/task$4.97$7.76$3.16$6.44 Min./task29.737.115.321.5 Output $$10$20$7.50$15 Carbon $2/$10 discount ends Nov 30Neon: +1 point, $1.32 less per task
  1. Four models in seven weeksArgon on September 30, Carbon on October 28, Hydrogen on November 12, Neon on November 18. Neon is adapted from Carbon.
  2. A small step in the indexNeon at xhigh scores 61, one point above Carbon at high. It ranks third among model families, behind GPT-6.2 Astra and Claude Fable 5.2.
  3. A bigger step in cost and timeOne index task costs $6.44 instead of $7.76 and takes 21.5 instead of 37.1 minutes. Neon’s tokens are 25% cheaper than Carbon’s standard price.
  4. What to watchUntil November 30, Carbon still sells at a discounted $2/$10. Google’s own table compares Neon with Claude Opus 5.5 and GPT-6.2 Sol, not with the index leaders.

Source: Artificial Analysis, Gemini 4 Neon model page and article, accessed Nov 19, 2026 (Intelligence Index v5.0.1, cost and time per task at standard prices, release dates); Google, „Introducing Gemini 4 Neon“, Nov 18, 2026; Gemini 4 Neon model card. Output price in US$ per million tokens, standard rates. Redrawn by Thorsten Meyer AI.

Graphic: Powered by Thorsten Meyer AI

The Trade-Off Between Score and Cost

The release shifts attention from a headline score to the cost and time needed to complete a defined set of tasks. At high, Neon matches Carbon’s score in the cited comparison for less money; at xhigh, it gains one point over Carbon-high while taking less time and costing less at standard prices. For developers choosing a model, those differences could affect how they allocate effort settings across routine and demanding work. The index results are benchmarks, however, and do not establish that Neon will be cheaper or faster for every application.

The setting choice itself can change the economics. Artificial Analysis lists Neon at 60 points and $3.74 per index task at high, compared with 61 and $6.44 at xhigh. Medium scores 55 for $1.90, while low scores 44 for $0.75. That makes testing the effort level against a real task an important part of evaluating the model, rather than assuming xhigh is the best default. These are reported benchmark costs, not guarantees of production spending.

Results also vary by test. Neon xhigh scores 63.6% on Terminal-Bench 4.0, below Carbon-high at 65.2%, according to the independent figures supplied. On AA-Omniscience, Neon’s reported hallucination rate is 14.4%, compared with Carbon’s 11.4%. Google separately claims Neon has its “lowest fabrication rate yet” in internal Antigravity testing, but that is a different measure and cannot be verified from the independent figures provided.

Philips 5500 Series Fully Automatic Espresso Machine with LatteGo Milk System

Philips 5500 Series Fully Automatic Espresso Machine with LatteGo Milk System

De'Longhi Magnifica Evo Fully Automatic Espresso Machine with Milk Frother, Silver

De'Longhi Magnifica Evo Fully Automatic Espresso Machine with Milk Frother, Silver

Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, IPX4, Black

Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, IPX4, Black

As an affiliate, we earn on qualifying purchases.

A Fourth Gemini Model in Seven Weeks

Neon is Google’s fourth model in about seven weeks, following Gemini 4 Argon on September 30, Carbon on October 28 and Hydrogen on November 12. Google positions the three current models for different uses: Carbon for hard, long-horizon projects, Hydrogen for fast everyday execution, and Neon between them. Its model card says Neon is adapted from Gemini 4 Carbon; for architecture, training data and hardware, it directs readers to Carbon’s documentation.

Google’s evaluation-methodology document compares Neon with Claude Opus 5.5 and GPT-6.2 Sol across 19 rows, marking Neon best in 16. Google itself shows Claude Opus ahead on Terminal-Bench 4.1 and FrontierCode 2.0, and GPT-6.2 Sol ahead on Chartography. Artificial Analysis ranks Neon third among model families on its Intelligence Index, behind GPT-6.2 Astra and Claude Fable 5.2, which are not included in Google’s table. Because the evaluations and comparison sets differ, their results should not be treated as directly interchangeable.

On Artificial Analysis’s Coding Agent Index, Neon scores 58.6, close to Claude Fable 5.2 at 58.5 and Claude Opus 5.5 at 58.2. The reported run costs are $6.86 for Neon, $15.47 for Fable and $12.61 for Opus. Those results offer a separate comparison focused on coding-agent tasks; they do not show that Neon leads across all coding work.

„“While Carbon tackles your hardest, long-horizon projects and Hydrogen handles fast everyday execution, Neon balances depth and efficiency.”“

— Google’s launch post

Benchmark Gaps and Open Questions

The benchmark record does not settle how Neon performs across all workloads. Google’s table and Artificial Analysis’s index use different evaluations, and Google’s table omits Claude Fable 5.2 and GPT-6.2 Astra, the two models ranked ahead of Neon by Artificial Analysis. The source material does not explain why those models were excluded, so the gap should not be treated as evidence of the models‘ performance in Google’s tests.

It is also unclear how the reported index costs and completion times translate to a particular product, prompt, workload or usage pattern. Neon’s higher output-token use can reduce its price advantage, and Carbon’s temporary discount changes the price comparison through November 30. Google’s internal fabrication-rate claim uses a separate test from the independent hallucination measure, so the two figures cannot establish which model is more reliable under the same conditions.

Testing Neon Against Real Workloads

The immediate next step for teams considering Neon is to compare its settings on their own tasks, tracking quality, completion time, token use and total cost rather than relying on a single index score. The cited figures make high a relevant starting point for that test: xhigh adds one point over high in Artificial Analysis’s reported Neon results, while raising task cost from $3.74 to $6.44. Results will depend on the workload and the prices in effect.

Google has not provided, in the source material, a reason for the missing models in its comparison table or a date for a broader side-by-side evaluation. Fairwind Program members eligible for the cyber-defense version were told to expect access over the next few days from the November 18 announcement. Further independent testing and fuller disclosures about evaluation methods would help clarify how Neon compares across coding, factual reliability and other tasks.

Key Questions

What is Gemini 4 Neon?

Gemini 4 Neon is a Google model released November 18, 2026. Google describes it as a balance between Carbon, intended for difficult long-horizon projects, and Hydrogen, aimed at fast everyday execution. Its model card says Neon is adapted from Carbon.

Is Neon smarter than Gemini 4 Carbon?

In the cited Artificial Analysis comparison, Neon at xhigh scores 61, one point above Carbon at high. Neon at high scores 60, matching Carbon’s cited high score. These benchmark results do not establish which model performs better on every task.

Why does the Google benchmark table need more context?

Google’s 19-row table compares Neon with Claude Opus 5.5 and GPT-6.2 Sol, but does not include Claude Fable 5.2 or GPT-6.2 Astra, which rank ahead of Neon in Artificial Analysis’s index. The source material does not explain the omission, and the separate evaluations may not be directly comparable.

Does Neon cost less than Carbon?

Neon’s standard list prices are $3 per million input tokens and $15 per million output tokens, below Carbon’s standard $4 and $20 rates. Carbon is discounted to $2 and $10 through November 30, according to the source material. Actual task costs depend on token use and workload.

Where can people use Gemini 4 Neon?

Google lists availability through the Gemini API in AI Studio, Gemini for business, Google Antigravity for paid AI Pro and Ultra subscribers, and the Gemini app for Pro and Ultra subscribers. A cyber-defense version is being offered to eligible Fairwind Program members.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

The AI Sales Test That Came Down to Reading One More File

The best AI agents won by following a document trail, finding a buried competitive fact and completing the €55,000 sale their analysis had earned.

Typesafe AI Raises $870M At $7.5B

Typesafe AI says it raised $870 million in a Series A led by Andreessen Horowitz, valuing the company at $7.5 billion.

IBM’s On-Prem AI Outlook: Open Weights Are Close, Scale Matters

IBM’s CEO says open-weight AI can approach frontier quality for less. Benchmarks and hardware costs show why the on-prem case depends on workload and scale.

The Machine Economy — Capital-Heavy, Human-Light, Trading With Itself

Analysis of the emerging machine economy where AI-driven firms operate with minimal human labor, trading mainly with each other, and its potential impacts.