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

🔍 Read the full analysis: How To Read Google’s Gemini 4 Neon Performance Table on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

  • 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, 2026, positioning it as a balance of capability and efficiency. Artificial Analysis measured a one-point advantage over Gemini 4 Carbon at the highest settings it compared, while Neon completed an index task faster and at lower standard-price cost. Google’s table shows Neon ahead in 16 of 19 rows against two selected rivals, but omits the models leading Artificial Analysis’s overall rankings.

Google released Gemini 4 Neon on November 18, describing it as a model that balances depth and efficiency. In Artificial Analysis’s comparison, Neon scored 61 on the Intelligence Index at xhigh, one point above Gemini 4 Carbon at high, while an index task cost less and took less time at standard list prices. The result is a modest score difference, not evidence that Neon leads every benchmark or task.

Artificial Analysis lists an index task at $6.44 and 21.5 minutes for Neon at xhigh, compared with $7.76 and 37.1 minutes for Carbon at high. At the same high setting, Neon scored 60 and cost $3.74 per index task. The figures describe the provider’s benchmark and cost methodology; they do not guarantee the same price or speed for every user workload.

The effort setting changes the trade-off. Artificial Analysis reports Neon at 61 points and $6.44 per task at xhigh; at the default high setting, it scores 60 for $3.74. Medium scores 55 at $1.90, while low scores 44 at $0.75. The figures suggest users should compare settings for their own tasks rather than treating the model name alone as a performance rating.

Google’s evaluation table contains 19 rows and marks Neon best in 16 against Claude Opus 5.5 and GPT-6.2 Sol. Google’s own table shows exceptions: Opus 5.5 leads Neon on Terminal-bench 4.1 and FrontierCode 2.0, while GPT-6.2 Sol leads on Chartography. That comparison set does not include Claude Fable 5.2 or GPT-6.2 Astra, which rank ahead of Neon on Artificial Analysis’s Intelligence Index.

Artificial Analysis’s Coding Agent Index places Neon at 58.6, Claude Fable 5.2 at 58.5 and Claude Opus 5.5 at 58.2. It estimates a coding-agent task at $6.86 for Neon, versus $15.47 for Fable 5.2 and $12.61 for Opus 5.5. These are results for that index’s test and pricing assumptions, not universal costs for coding work.

At a glance
reportWhen: Released November 18, 2026; performance…
The developmentGoogle released Gemini 4 Neon on November 18, and new independent comparisons show a small score gain over Carbon alongside lower task costs and shorter 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

Neon’s Score, Speed and Cost Trade-Off

The practical question is not simply whether Neon beats Carbon. The reported results point to a small measured gain at the highest compared settings, paired with a larger difference in the index’s task time and cost estimates. For teams running repeated model tasks, those costs and completion times can matter alongside benchmark scores.

But the results do not establish a universal winner. Neon’s setting affects both score and cost, and it trails Carbon on at least one reported test: Terminal-Bench 4.0, where Neon at xhigh scored 63.6% and Carbon at high scored 65.2%. The cost figures also use standard list prices; a temporary Carbon discount changes the comparison.

Google’s 16-of-19 result is informative only with its comparison set attached. Readers who treat it as proof that Neon beats the leading models across the field would be going beyond the evidence presented. Independent index results put GPT-6.2 Astra and Claude Fable 5.2 ahead overall, though different test suites can produce different orderings.

Products Worth Considering

FLEXISPOT Oversized Outdoor Rocking Chair with Thick Cushion, Grey

FLEXISPOT Oversized Outdoor Rocking Chair with Thick Cushion, Grey

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite

Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite

As an affiliate, we earn on qualifying purchases.

Google’s Rapid Gemini 4 Releases

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

Google says Neon accepts text, images, audio and video, produces text, and has a one-million-token context window. The model card gives a March 2026 knowledge cutoff and four effort levels, with high as the default. Google lists access through the Gemini API in AI Studio and several paid Gemini services. A cyber-defense version is being made available to trusted defenders through the company’s Fairwind Program, with eligible members expected to receive access over several days, according to Google.

Artificial Analysis summarizes its comparison as a small Intelligence Index gain over Carbon alongside a roughly 17% lower cost per task. It attributes the cost difference to Neon’s lower token prices and faster output, partly offset by Neon using about 15% more output tokens per task. Its listed standard rates are $3 per million input tokens and $15 per million output tokens for Neon, compared with Carbon’s standard $4 and $20. Carbon’s rates were discounted to $2 and $10 through November 30, according to the supplied report.

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

— Google’s description of the Gemini 4 model lineup

Limits of the Published Comparisons

The available figures do not establish how Neon performs across every real-world workload. Artificial Analysis’s scores and task-cost estimates apply to its tests and methodology, while Google’s benchmark table uses a separate evaluation set. The supplied material does not provide enough detail to reconcile every test difference or to verify Google’s internal fabrication-rate claim independently.

Google’s table also leaves out Claude Fable 5.2 and GPT-6.2 Astra, the two models ahead of Neon in the Artificial Analysis Intelligence Index. The comparisons therefore cannot show how Neon fares against those models on all 19 of Google’s listed tests. Nor do the published list prices alone establish what a particular customer will pay: usage, settings, discounts and service terms can affect actual costs.

The source material gives a temporary Carbon discount through November 30 but does not say whether that offer will be extended. It also does not provide a like-for-like comparison of all models at identical effort settings across every cited test. Readers should keep those limitations in view when interpreting claims of a win or a price advantage.

Testing Settings and Access

Readers evaluating Neon can begin by matching the effort level to the task and comparing quality, completion time and actual usage cost on their own workloads. Artificial Analysis’s figures show that moving from high to xhigh adds one index point while increasing estimated task cost from $3.74 to $6.44. That is a benchmark-based starting point, not a forecast for any individual deployment.

Google lists Neon as available through the Gemini API and named Gemini services, with a separate cyber-defense version for eligible Fairwind Program members. The next useful evidence will be broader, comparable testing against the omitted leading models, along with data from users’ real tasks and clarification of pricing after Carbon’s stated discount period. The supplied information does not give a date for those comparisons or any further benchmark release.

Key Questions

What is the main result in the Gemini 4 Neon performance data?

Artificial Analysis gave Neon at xhigh an Intelligence Index score of 61, one point above Carbon at high. It also reported a lower estimated task cost and shorter task time for Neon in that comparison.

Does Google’s table show Neon beating every model it compares?

No. Google marks Neon best in 16 of 19 rows, but its table shows Claude Opus 5.5 ahead on two tests and GPT-6.2 Sol ahead on one. The table does not include Claude Fable 5.2 or GPT-6.2 Astra.

Why do Neon’s cost figures vary by setting?

Artificial Analysis reports different scores and estimated costs at each effort level. Its figures range from $0.75 per index task at low to $6.44 at xhigh, with the high setting at $3.74. These estimates use its benchmark and standard list-price assumptions.

Is Neon cheaper than Carbon?

Neon’s listed standard token prices are lower: $3 per million input tokens and $15 per million output tokens, compared with Carbon’s standard $4 and $20. The supplied report says Carbon was discounted through November 30, 2026, so the temporary offer narrows the price difference during that period.

What remains unverified about Google’s claims?

Google’s reported “lowest fabrication rate” refers to internal Antigravity testing; the supplied material says external readers cannot check that measurement. Independent benchmark results also do not cover every model, task or deployment cost.

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

Mistral Forge: Owning the Model, Not Just Renting the API

Mistral’s Forge offers organizations the ability to own and operate their own AI models, moving beyond API rentals. This impacts enterprise AI strategies and sovereignty.

How AI Is Making Gaming Smarter: 9 Key Innovations In 2026

Nine key AI-driven innovations are making video games smarter in 2026, enhancing realism, responsiveness, and player experience.

Could Noise Cancelling Headphones Help With AI Work? 9 Picks For 2027

A 2027 buyer’s guide ranks nine headphone listings for work, calls and focus, while noting that the recommendations are not based on hands-on testing.

Choosing An AI Video Generator: 4 Best Options For 2027

A comparison of four AI video generators based on their descriptions, including the workflows named and the product details still missing.