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

🔍 Read the full analysis: A Practical Guide To Understanding Gemini 4 Neon’s Numbers on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

Google released Gemini 4 Neon on November 18, 2026, positioning it as a balance of capability and efficiency. Artificial Analysis scores Neon one point above Gemini 4 Carbon at the models’ highest compared settings, while reporting lower cost and shorter completion time for an index task. Results vary by setting and benchmark, and Google’s published comparison omits two models that lead the independent index.

Google released Gemini 4 Neon on November 18, describing it as a model built for fast task completion. Figures from benchmark publisher Artificial Analysis show Neon scoring one point above Gemini 4 Carbon at the compared high-end settings, while costing less and taking less time on the index’s tasks. The results do not make Neon the overall leader, and the settings used affect both its score and price.

On the Artificial Analysis Intelligence Index, Neon at its xhigh setting scored 61, compared with Carbon at high with 60. Artificial Analysis ranked Neon third among model families, behind GPT-6.2 Astra and Claude Fable 5.2. At standard list prices, the index’s representative task cost $6.44 with Neon at xhigh, against $7.76 with Carbon at high, and took 21.5 minutes rather than 37.1 minutes.

The setting changes the comparison. At high, Neon’s default, it scored 60 and cost $3.74 per index task, according to Artificial Analysis. Medium scored 55 at $1.90 and low scored 44 at $0.75. Moving from high to xhigh added one index point but raised the reported task cost to $6.44 and time to 21.5 minutes. These figures describe the index’s tasks, not a guaranteed price or completion time for every user’s work.

Google lists Neon at $3 per million input tokens and $15 per million output tokens. Carbon’s standard rates are $4 and $20, though Google’s reported Carbon discount of $2 and $10 per million tokens runs through November 30. Artificial Analysis attributes Neon’s lower standard-price task cost partly to its lower token rates and faster output; it also reports that Neon used about 15% more output tokens per task than Carbon, which reduces the savings. Cached input is listed at $0.15 per million tokens, a 95% discount from Neon’s standard input rate.

At a glance
reportWhen: Released November 18, 2026; benchmark f…
The developmentGoogle released Gemini 4 Neon, a model that independent benchmark data scores slightly above Carbon while showing lower task costs and faster 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

Neon’s Cost Depends on Its Setting

For buyers, the figures point to a cost-and-speed trade-off, not a large jump in measured intelligence. The benchmark’s one-point difference between Neon xhigh and Carbon high is small, while reported task cost and time differ more. At Neon’s default high setting, the reported cost is lower still, but its score is 60 rather than 61. Teams comparing models should match settings and task requirements instead of treating one headline score as a complete ranking.

Neon also comes close to competing coding models in Artificial Analysis’s Coding Agent Index: Neon scored 58.6, Claude Fable 5.2 scored 58.5, and Claude Opus 5.5 scored 58.2. The reported cost per task was $6.86 for Neon, compared with $15.47 for Fable and $12.61 for Opus. Those results may interest developers weighing operating costs, but they do not establish that every coding workload will be cheaper or perform equally well.

Performance varies across individual tests. Neon xhigh scored 63.6% on Terminal-Bench 4.0, below Carbon high at 65.2%. On Artificial Analysis’s AA-Omniscience test, Neon’s reported hallucination rate was 14.4%, against Carbon’s 11.4%. Google separately said Neon had its lowest fabrication rate yet in internal Antigravity testing; that is a company-reported result from a different evaluation and cannot be directly equated with the independent test.

Products Worth Considering

As an affiliate, we earn on qualifying purchases.

A Fourth Gemini Model in Seven Weeks

Neon is Google’s fourth Gemini 4 model in about seven weeks, following Argon on September 30, Carbon on October 28 and Hydrogen on November 12. Google describes Carbon as suited to difficult, long-horizon projects and Hydrogen as designed for fast everyday execution, with Neon positioned between them. Its model card says Neon is adapted from Carbon and directs readers to Carbon’s documentation for information on architecture, training data and hardware.

Google’s model card and launch information list inputs for text, images, audio and video, with text output and a one-million-token context window. Output is up to 64,000 tokens by default, with up to one million available through background long decode. The stated knowledge cutoff is March 2026, and users can choose among low, medium, high and xhigh effort levels; high is the default.

Google’s evaluation document reports Neon as best in 16 of 19 rows against Claude Opus 5.5 and GPT-6.2 Sol. Google’s own table also shows rivals ahead in three rows: Claude Opus 5.5 on Terminal-bench 4.1 and FrontierCode 2.0, and GPT-6.2 Sol on Chartography. The set does not include Claude Fable 5.2 or GPT-6.2 Astra, which lead Neon in the Artificial Analysis Intelligence Index. The two scorecards use different comparisons, so their results should not be treated as a single shared ranking.

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

— Google, describing the Gemini 4 model lineup

Benchmark Gaps and Open Questions

The available figures do not settle which model is best for a particular user. The index scores are tied to specific settings and tests, and results on one benchmark do not establish performance across all tasks. Google’s comparison table uses a different and narrower set of competitors than Artificial Analysis’s index, so readers cannot use the vendor table alone to compare Neon with the two index leaders.

It is also unclear how closely the published task costs will match typical production use. Token use, chosen effort level, prompt and output length, caching, workload and access terms can change a bill. Artificial Analysis reports its figures at standard list prices; Carbon’s temporary discount changes the price comparison through November 30. The source material does not provide a complete account of how every benchmark result was independently reproduced.

Google’s claim about its lowest fabrication rate comes from internal Antigravity testing, while Artificial Analysis’s hallucination figure comes from AA-Omniscience. The tests are different, and the internal result cannot be checked from the details provided. Google also says Gemini 4 Neon for cyber will reach eligible defenders in its Fairwind Program “over the next few days”; access timing and broader availability beyond that statement remain unspecified.

Availability and Price Changes

Google says Neon is available through the Gemini API in Google AI Studio and Gemini for business, as well as Google Antigravity and the Gemini app for paid AI Pro and Ultra subscribers. The company said the cyber-defense version would become available to eligible Fairwind Program members over the following days. The source material does not give a later date for wider access to that version.

For teams evaluating Neon, the next useful step is to test their own tasks at the effort levels they expect to use and record both quality and token use. The published index figures suggest comparing high with xhigh before paying for the additional effort: xhigh added one point in the cited score but raised the reported task cost. Buyers comparing Carbon should also account for the stated discount ending November 30, rather than relying only on standard list prices.

Further independent results may clarify how Neon performs across more workloads and against the models absent from Google’s table. Until then, the confirmed picture is narrower: Neon is a Carbon-derived model with a modestly higher index score at the compared high settings, alongside lower reported task costs and shorter completion times in Artificial Analysis’s evaluation.

Key Questions

Is Gemini 4 Neon the top-ranked model?

No. Artificial Analysis ranked Neon third among model families on its Intelligence Index, behind GPT-6.2 Astra and Claude Fable 5.2. Rankings vary by benchmark and setting.

What is Neon’s default effort setting?

Google lists high as the default. Artificial Analysis reports a score of 60 and a cost of $3.74 per index task for Neon at high, compared with a score of 61 and $6.44 at xhigh.

Is Neon cheaper than Gemini 4 Carbon?

At standard list prices, Neon’s input and output token rates are lower than Carbon’s: $3 and $15 per million tokens, versus $4 and $20. Artificial Analysis also reports lower cost per index task for Neon xhigh than Carbon high, but Carbon’s temporary discount through November 30 makes the current price comparison different.

Where can users access Gemini 4 Neon?

Google lists availability through the Gemini API in Google AI Studio, Gemini for business, and Google Antigravity and the Gemini app for paid AI Pro and Ultra subscribers.

What remains uncertain about the results?

The reported benchmark results do not establish how Neon will perform or cost on every real-world workload. Google’s internal fabrication-rate claim uses a different test from Artificial Analysis’s measurement, and Google’s vendor comparison omits two models that lead the independent index.

Source: ThorstenMeyerAI.com

You May Also Like

Huang’s Offer To Return, Nvidia’s Rejection, And The AI Stakes

Nvidia says it does not intend to reinstate Jensen Huang after his return offer. Congressional interviews are scheduled for Dec. 1 and 2.

What’s Happening To CVEs? A Data Check On AI And Patch Gaps

A data review found 13,934 CVE records published by Oct. 25, 2026, while Anthropic reported only 516 of 6,157 disclosed flaws patched.

Anthropic’s AI Work Gets A New Question As Three Claude Code Members Leave

Analytics India Magazine reports three people linked to Claude Code left Anthropic within 10 days; their roles and reasons remain unknown.

Mapquest Surges In Global Coverage

Mapquest’s coverage has surged globally, with 14 mentions in recent monitoring, marking a major expansion in its mapping services.