🔍 Read the full analysis: Gemini 4 Neon Assessed: Modest Results, Lower Costs, And An Incomplete View on ThorstenMeyerAI.com
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
TL;DR
Google released Gemini 4 Neon on November 18, describing it as a model that balances depth and efficiency. Artificial Analysis measured a one-point gain over Gemini 4 Carbon at the highest effort setting, alongside lower cost and shorter completion time; results vary by setting and test. Google’s published comparison table also excludes two models that lead the independent index, limiting what it shows about Neon’s standing against the top-ranked systems.
Google released Gemini 4 Neon on November 18, presenting it as a model that balances capability and efficiency. In figures from Artificial Analysis, Neon scored 61 on the Intelligence Index at its xhigh setting, one point above Gemini 4 Carbon at high, while a measured index task cost $6.44 and took 21.5 minutes, compared with $7.76 and 37.1 minutes for Carbon at its standard price. The results indicate a modest score gain alongside lower cost and faster completion, not a lead across every test.
The independent comparison is not a like-for-like effort-setting test: Neon’s 61-point result uses xhigh, while Carbon’s 60-point result uses high. At the same high setting, Neon scores 60 and costs $3.74 per index task, according to Artificial Analysis. The index task-cost figures use standard list prices and reflect the model settings shown; they should not be treated as a universal cost for every user or workload.
Neon’s listed prices are $3 per million input tokens and $15 per million output tokens, below Carbon’s standard rates of $4 and $20. However, Carbon is discounted to $2 and $10 through November 30, according to the source report. Artificial Analysis also reports that Neon generates about 110 output tokens per second, against about 55 for Carbon, while using roughly 15% more output tokens per task. That added usage offsets some of the lower per-token price.
Performance depends on the requested effort. Artificial Analysis reports Neon at 61 points and $6.44 per task at xhigh, 60 and $3.74 at high, 55 and $1.90 at medium, and 44 and $0.75 at low. Going from high to xhigh therefore adds one index point in these measurements while increasing task cost. On the Coding Agent Index, Neon scored 58.6, close to Claude Fable 5.2 at 58.5 and Claude Opus 5.5 at 58.2; the reported task costs were $6.86, $15.47 and $12.61, respectively.
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.
- 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.
- 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.
- 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.
- 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 Cost of Higher Effort
For developers and businesses choosing a model, the release’s clearest reported difference is not a dramatic jump in benchmark score but a trade-off between effort, price and speed. Artificial Analysis’s figures suggest Neon can complete its index tasks faster than Carbon and cost less at standard rates. At the same time, increasing Neon from high to xhigh costs substantially more for a one-point increase on that index. Buyers may need to test settings against their own tasks rather than assume the highest setting offers enough added value to justify its price.
The comparison also shows why a single overall score is incomplete. Neon’s coding-agent score is close to two Claude models, but on Terminal-Bench 4.0 it scored 63.6%, below Carbon’s 65.2%. On the AA-Omniscience test, Artificial Analysis measured a 14.4% hallucination rate for Neon at its highest setting, compared with 11.4% for Carbon. These results make the release relevant to teams weighing cost and throughput, but they do not establish that Neon is the best choice across coding, factual accuracy or other uses.
Products Worth Considering

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

Bose SoundTouch 30 Wireless Speaker, Black

Apple Watch SE 3 GPS + Cellular 40mm Smartwatch
As an affiliate, we earn on qualifying purchases.
A Fourth Model in Seven Weeks
Neon is the fourth Gemini 4 model Google has 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 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 architecture, training data and hardware details.
Google lists support for text, images, audio and video inputs, text output and a one-million-token context window. The default output limit is 64,000 tokens, with up to one million available through background long decode. The model’s stated knowledge cutoff is March 2026, and its effort settings range from low to xhigh, with high as the default. Google says Neon is available through the Gemini API in AI Studio and several paid Gemini products. A cyber-defense version is being made available to trusted defenders in the company’s Fairwind Program.
Google’s evaluation document compares Neon with Claude Opus 5.5 and GPT-6.2 Sol across 19 rows, marking Neon best in 16. The same table shows Claude Opus ahead on Terminal-Bench 4.1 and FrontierCode 2.0, and GPT-6.2 Sol ahead on Chartography. The comparison does not include Claude Fable 5.2 or GPT-6.2 Astra, which rank ahead of Neon in the Artificial Analysis Intelligence Index. The two sources therefore answer different questions and use different comparison sets.
“Frontier-level intelligence with fast end-to-end task completion.”
— Google, in its Gemini 4 Neon launch post
Limits of the Comparisons
The reported benchmarks do not settle how Neon will perform across real-world workloads. Artificial Analysis provides results for specific tests and settings, while Google’s evaluation table uses a different set of comparisons. Google’s statement that Neon had its “lowest fabrication rate yet” in internal Antigravity testing refers to an internal measurement that outside readers cannot independently check from the information provided.
The comparison against the index leaders is also incomplete in Google’s table: it does not include Claude Fable 5.2 or GPT-6.2 Astra. The source report says independent measurements place those models ahead of Neon on most of the same tests, but does not provide a complete row-by-row comparison here. It is also unclear from the material provided how results will change as prices, discounts, model versions or evaluation methods change. Carbon’s stated discount expires November 30, which affects the short-term price comparison.
Availability and Further Testing
Users can access Neon through the Gemini API in Google AI Studio and Google’s listed Gemini services, with some app and Antigravity access limited to Pro and Ultra subscribers. Google said eligible Fairwind Program members would receive access to the cyber-defense version over the following few days. The report does not give a later date for that rollout or specify wider availability for the cyber model.
For prospective users, the next useful evidence will be testing at the effort setting and price they expect to use, followed by comparisons on their own workloads. Independent results that include the omitted index leaders, alongside more detail on Google’s internal fabrication test, would make the model’s relative performance easier to judge. Until then, the available figures support a limited finding: Neon’s benchmark gain over Carbon is small, while reported task cost and completion time are lower in the stated comparison.
Key Questions
How much did Gemini 4 Neon improve over Carbon?
Artificial Analysis scored Neon at 61 at xhigh, one point above Carbon’s 60 at high. At the same high setting, Neon also scored 60, matching Carbon’s reported score in the comparison.
Is Gemini 4 Neon cheaper than Carbon?
Neon’s standard listed prices are $3 per million input tokens and $15 per million output tokens, compared with Carbon’s standard $4 and $20. Carbon is reported to be discounted to $2 and $10 through November 30, so its temporary prices are lower than Neon’s listed rates.
Does Neon outperform the other leading models?
The available evidence does not show that across the board. 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 Neon on the Artificial Analysis Intelligence Index.
Where can users access Gemini 4 Neon?
Google lists Neon in the Gemini API through AI Studio and in selected Gemini products. Access to Google Antigravity and the Gemini app is listed for paid Pro and Ultra subscribers; exact availability may depend on the product.
Source: ThorstenMeyerAI.com
