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

🔍 Read the full analysis: Gemini 4 Neon: A Small Step In Performance And A Smaller Bill on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

Google released Gemini 4 Neon on November 18, positioning it as a more efficient model adapted from Gemini 4 Carbon. Artificial Analysis scores Neon one point above Carbon at its highest effort setting and reports lower cost and shorter completion time on its index tasks; results vary by setting and benchmark.

Google released Gemini 4 Neon on November 18, describing it as a model that balances depth and efficiency. In figures published by independent evaluator Artificial Analysis, Neon scores 61 at its xhigh effort setting, one point above Gemini 4 Carbon at high, while an index task costs $6.44 rather than $7.76 and takes 21.5 rather than 37.1 minutes at those settings.

The comparison is not like-for-like on effort: Neon’s xhigh setting is being compared with Carbon’s high setting. At the same high setting, Artificial Analysis gives both models a score of 60; Neon’s index task costs $3.74, compared with $7.76 for Carbon at its standard price. The evaluator’s figures were accessed November 19, and task costs are based on standard list prices.

Google lists Neon at $3 per million input tokens and $15 per million output tokens, against Carbon’s standard $4 and $20. Carbon is discounted to $2 and $10 through November 30, according to the source material, making it temporarily cheaper on listed token rates. Artificial Analysis attributes Neon’s lower task cost to the lower standard prices and faster output, while noting that Neon uses about 15% more output tokens per task.

Neon has four effort levels, from low to xhigh, with high as the default. Artificial Analysis reports scores and costs of 44 and $0.75 at low, 55 and $1.90 at medium, 60 and $3.74 at high, and 61 and $6.44 at xhigh. The results show that choosing a higher effort level can affect cost as much as choosing between model versions.

At a glance
updateWhen: Released November 18, 2026; performance…
The developmentGoogle released Gemini 4 Neon, a model adapted from Gemini 4 Carbon that posts a small independent index-score gain alongside lower measured task costs and completion times.

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

Lower Costs, With Setting Trade-Offs

For teams running repeated model tasks, a lower cost per completed task can matter more than a small increase in a benchmark score. Artificial Analysis reports Neon’s xhigh index tasks take 21.5 minutes, compared with 37.1 minutes for Carbon at high, and cost less despite the difference in effort setting. At high, Neon matches Carbon’s score for less than half the reported task cost at Carbon’s standard price.

Those numbers are benchmark-specific, not a guarantee of savings in every application. The source reports that Neon is about 110 output tokens per second, versus about 55 for Carbon, but also uses more output tokens per task. Real costs and response times will depend on workload, usage patterns, effort setting and any applicable discounts. Users deciding whether to switch should test their own tasks rather than treating the index result as a universal measure.

Performance is mixed across evaluations. On the Artificial Analysis 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. But on Terminal-Bench 4.0, Neon scores 63.6%, below Carbon’s 65.2% at high. The results support a case for testing Neon on cost-sensitive coding work, not a claim that it leads on every task.

Samsung 65-Inch 4K OLED Smart TV with Dolby Atmos, 120Hz Refresh Rate, and Amber Protection Bundle

Samsung 65-Inch 4K OLED Smart TV with Dolby Atmos, 120Hz Refresh Rate, and Amber Protection Bundle

As an affiliate, we earn on qualifying purchases.

Neon’s Place in Google’s Lineup

Google describes Neon as sitting between Gemini 4 Carbon, for difficult long-horizon projects, and Gemini 4 Hydrogen, for fast everyday execution. Its model card says Neon is adapted from Carbon and points readers to Carbon’s documentation for architecture, training data and hardware details. That makes the release an efficiency-focused variation of an existing model family, rather than a wholly separate model according to the information provided.

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 lists support for text, images, audio and video as inputs, with text output and a one-million-token context window. The stated knowledge cutoff is March 2026. Availability includes the Gemini API in AI Studio and several Google business and subscription products; a cyber-defense version is being offered to eligible trusted defenders through Google’s Fairwind Program.

Google’s evaluation document reports Neon ahead in 16 of 19 listed rows, but its comparison set is limited to Claude Opus 5.5 and GPT-6.2 Sol. It does not include Claude Fable 5.2 or GPT-6.2 Astra, which the supplied Artificial Analysis results place ahead of Neon on the Intelligence Index. In Google’s own table, Opus leads Neon on Terminal-bench 4.1 and FrontierCode 2.0, while GPT-6.2 Sol leads on Chartography.

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

— Google, in its Gemini 4 Neon launch post

Benchmark Limits and Open Questions

The supplied material does not provide enough detail to establish how Neon performs across every customer workload, or whether the reported time and cost differences will hold for production use. Artificial Analysis’s comparison uses different effort settings for its headline Neon-versus-Carbon figures; its high-setting comparison is closer in effort and shows equal scores. The effects of workload, output length and billing conditions remain relevant.

Google’s evaluation table covers 19 rows but leaves out the two models identified in the supplied independent ranking as ahead of Neon on the Intelligence Index. The source does not provide a complete, directly matched set of results across all models and tests. Google’s claim that Neon had its “lowest fabrication rate yet” in internal Antigravity testing also uses a different measurement from the independent AA-Omniscience figures, so the results cannot be treated as equivalent. The internal test methodology and results are not independently verifiable from the information provided.

The source material does not state whether Google plans to revise prices, publish further evaluation details or expand access to the cyber-defense version after the initial rollout. Availability and commercial terms may also differ by product or account.

Testing Neon Against Real Workloads

The immediate next step for prospective users is to compare Neon with their current model on representative tasks, tracking quality, completion time and total cost at the effort levels they expect to use. The supplied figures show that moving from high to xhigh adds one index point but raises the reported task cost from $3.74 to $6.44, a trade-off that may not suit every workload.

Users weighing a switch should also account for Carbon’s stated discount through November 30, rather than comparing Neon’s standard prices only with Carbon’s undiscounted rates. Google’s launch materials list Neon across its API and selected business and paid subscription products. Eligible Fairwind Program members are expected to receive access to the cyber-defense version over the following few days, according to the supplied source.

Further independent measurements and more comparable tests against competing models would help clarify how much Neon’s cost and speed advantages carry over beyond the published index tasks. Until then, the available evidence supports a measured conclusion: Neon improves some efficiency metrics over Carbon, but the size of the benefit depends on settings, pricing and the task being run.

Key Questions

When did Google release Gemini 4 Neon?

Google released Gemini 4 Neon on November 18, 2026, according to the supplied source material.

Does Neon score higher than Gemini 4 Carbon?

On the Artificial Analysis Intelligence Index, Neon scores 61 at xhigh, one point above Carbon at high. At the same high setting, both score 60. The xhigh-versus-high comparison uses different effort levels.

Is Neon cheaper than Carbon?

Neon’s standard listed rates are $3 per million input tokens and $15 per million output tokens, below Carbon’s standard $4 and $20 rates. The source says Carbon is discounted to $2 and $10 through November 30, so Carbon’s temporary listed rates are lower during that period. Artificial Analysis reports a lower Neon cost per index task in its stated comparisons.

Does Neon lead every benchmark in the supplied results?

No. It is close to Claude models on the cited coding index, but scores below Carbon on Terminal-Bench 4.0. Google’s table also lists two tests where Claude Opus 5.5 leads and one where GPT-6.2 Sol leads.

Where is Gemini 4 Neon available?

The supplied material lists the Gemini API in Google AI Studio, Gemini for business, Google Antigravity for paid Google AI Pro and Ultra subscribers, and the Gemini app for Pro and Ultra subscribers. Eligible trusted defenders are also due access to a cyber-defense version through Google’s Fairwind Program.

Source: ThorstenMeyerAI.com

You May Also Like

Smart Charging Algorithms: Balancing Grid Demand and Fleet Needs

Keen to optimize EV charging efficiency and sustainability? Discover how smart algorithms balance grid demand and fleet needs seamlessly.

Anthropic’s New Rules Target Cruel Treatment Of Claude

The Verge reports that Anthropic has banned abusive or cruel behavior toward Claude, but the rule’s scope, timing and enforcement are not detailed.

Why Chrome’s Eighth Zero-Day Of 2026 Puts XML In Focus

Google patched a high-severity XML use-after-free in Chrome after confirming in-the-wild exploitation. Here are the fixed versions and what remains unknown.

The Future Of Open Source AI? Huawei’s 505B openPangu And Its Open Code

Huawei has open-sourced its 505B openPangu AI model, releasing weights and code. Details on licensing, performance, and usage remain unclear.