What The Gemini 4 Neon Comparison Reveals About Cost And Performance
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: What The Gemini 4 Neon Comparison Reveals About Cost And Performance 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. Artificial Analysis measured a one-point score gain over Gemini 4 Carbon at the highest tested settings, alongside lower cost and shorter task times; results vary by setting and benchmark. Google’s comparison table does not include the two models that lead the independent index.

Google released Gemini 4 Neon on November 18, presenting it as a balance of depth and efficiency; independent testing found it scored one point above Gemini 4 Carbon at the models’ highest listed settings, while costing less and completing the measured tasks faster. Artificial Analysis gave Neon a score of 61 at xhigh, versus Carbon’s 60 at high, and estimated per-task costs of $6.44 and $7.76, respectively.

The figures come from the Artificial Analysis Intelligence Index, which reports cost and elapsed time for a set of tasks. Its results put Neon at 21.5 minutes per task, compared with Carbon’s 37.1 minutes. These are benchmark measurements, not a guarantee that Neon will finish every real-world request faster or more cheaply.

Neon’s effort setting changes the balance. At the default high setting, it scores 60 and costs $3.74 per index task; xhigh raises the score to 61 but the cost to $6.44. At medium, the figures are 55 and $1.90, and at low they are 44 and $0.75. Artificial Analysis says Neon generates about 110 output tokens per second, roughly twice Carbon’s rate, but uses about 15% more output tokens per task.

Google lists Neon at $3 per million input tokens and $15 per million output tokens, compared with Carbon’s standard rates of $4 and $20. Carbon is temporarily discounted to $2 and $10 through November 30, according to the source material, so Neon is not cheaper than Carbon at those promotional prices. Cached input for Neon costs $0.15 per million tokens, a 95% discount from its standard input rate.

At a glance
reportWhen: Released November 18, 2026; figures acc…
The developmentGoogle launched Gemini 4 Neon, a model adapted from Carbon that posts a marginally higher independent intelligence score while costing less per measured task.

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

Where Neon Saves Time and Cost

For teams paying to run long or repeated model tasks, cost per completed task may matter more than a small change in an aggregate intelligence score. The independent figures suggest Neon can reduce both expense and waiting time relative to Carbon at the settings compared, but they also show a trade-off: moving from high to xhigh adds one index point while increasing the measured task cost by $2.70.

The results do not establish a universal winner. On the Artificial Analysis Coding Agent Index, Neon scored 58.6, close to Claude Fable 5.2 at 58.5 and Claude Opus 5.5 at 58.2. Its measured coding task cost was $6.86, versus $15.47 for Fable 5.2 and $12.61 for Opus 5.5. Yet on Terminal-Bench 4.0, Neon at xhigh scored 63.6%, below Carbon at high, at 65.2%. Buyers need to match tests and effort settings to their own workloads rather than treating one ranking as decisive.

Products Worth Considering

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

As an affiliate, we earn on qualifying purchases.

A Rapid Gemini 4 Release Cadence

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, according to the supplied report. Google describes Carbon as suited to difficult, long-horizon projects and Hydrogen to fast everyday execution, positioning Neon between them.

Google’s model card says Neon is adapted from Gemini 4 Carbon; its documentation directs readers to Carbon’s materials for architecture, training data and hardware. The model accepts text, images, audio and video, produces text, and has a one-million-token context window. Google lists a March 2026 knowledge cutoff and four effort levels from low to xhigh, with high as the default.

Google’s published evaluation table reports Neon ahead in 16 of 19 rows against Claude Opus 5.5 and GPT-6.2 Sol. The company itself lists rivals ahead on Terminal-bench 4.1 and FrontierCode 2.0 for Opus, and Chartography for GPT-6.2 Sol. That table’s comparison set excludes Claude Fable 5.2 and GPT-6.2 Astra, which lead the Artificial Analysis Intelligence Index. The independent index and Google’s table use distinct evaluations, so their results should not be treated as directly interchangeable.

“frontier-level intelligence with fast end-to-end task completion”

— Google DeepMind

Benchmark Gaps and Open Questions

The supplied material does not provide enough detail to establish how representative the index tasks are of each customer’s use, or whether the reported cost and speed differences will hold across other prompts, tools and operating conditions. Results also compare different effort settings for Neon and Carbon in key headline figures: xhigh for Neon and high for Carbon. Neon’s high-setting score matches Carbon’s high-setting score, but the source does not give a matching Carbon cost and time comparison at every Neon setting.

Some claims cannot be independently checked from the reported figures. Google says Neon had its lowest fabrication rate in internal Antigravity testing, but that is a different measurement from Artificial Analysis’s AA-Omniscience test. On that independent test, Neon had a 14.4% hallucination rate, compared with Carbon’s 11.4%; scores and settings differ across the other models as well. Google’s launch statement does not make the internal result independently verifiable.

The source report also discloses that it was researched and drafted with AI assistance, including Anthropic’s Claude, and says Google, Anthropic, OpenAI and Artificial Analysis did not review or sponsor it. Readers should consult the linked original model and benchmark pages for full evaluation methods. The material provided ends partway through its discussion of Google’s comparison table, leaving the complete explanation of its methodology and other caveats unavailable here.

Availability and Pricing Checks

Google says Neon is available through the Gemini API in Google AI Studio, Gemini for business, and the Gemini app for Pro and Ultra subscribers. Paid Google AI Pro and Ultra subscribers can also access it through Google Antigravity. The source says a cyber-defense version is being offered to trusted defenders in Google’s Fairwind Program, with eligible members expected to receive access over the following few days.

For prospective users, the next practical step is to test Neon at the effort level and workload they actually need, then compare task quality, latency and token use against alternatives. The Carbon discount is listed as running through November 30, 2026, making the price comparison time-sensitive. The supplied material gives no further release milestones or dates for additional benchmark results, so changes to pricing, access and independent rankings remain to be confirmed.

Key Questions

How much better did Gemini 4 Neon score than Carbon?

Artificial Analysis scored Neon at 61 at xhigh and Carbon at 60 at high on its Intelligence Index. The one-point difference is not a like-for-like comparison of settings. At high, Neon scored 60, matching Carbon’s reported high score.

Is Neon cheaper than Carbon?

At standard list prices, yes: Google lists Neon at $3 per million input tokens and $15 per million output tokens, against Carbon’s $4 and $20. Carbon is discounted to $2 and $10 through November 30, 2026, according to the source material. Artificial Analysis also measured lower Neon cost per index task in its stated comparison.

Which model is best for coding?

The reported results do not identify a clear winner across coding tests. Neon, Claude Fable 5.2 and Claude Opus 5.5 scored within half a point on the Artificial Analysis Coding Agent Index; Neon scored below Carbon on Terminal-Bench 4.0 in the settings cited.

Where can people access Gemini 4 Neon?

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

Do the available benchmarks prove Neon is better than leading rival models?

No. Google’s table compares Neon with Claude Opus 5.5 and GPT-6.2 Sol but omits Claude Fable 5.2 and GPT-6.2 Astra, which lead the cited Artificial Analysis index. Benchmark results vary by task, settings and evaluation method.

Source: ThorstenMeyerAI.com

You May Also Like

Squeak 6.1

Squeak 6.1, the latest version of the Smalltalk environment, has been officially released, introducing performance enhancements and new features for developers.

Google Arts & Culture Turns 15 — And Gives Its App A Makeover

Google is rolling out a redesigned Arts & Culture app with conversational search, two AI experiments and 27 new city guide destinations.

A New Way To Play With Prompts: OpenAI’s Chromatic Codex Edition

OpenAI began selling a $200 ModRetro Chromatic bundled with a programmable Codex Cartridge. The page showed a November 19 restock date on Nov. 10.

From Concept To Screen: The Impact Of AI And Seedance 2.5 On Modern Film

ByteDance Seed announced Seedance 2.5, an AI model claiming to produce 30-second continuous video sequences, but technical details and availability remain unverified.