TL;DR
Google released Gemini 4 Neon on November 18. Artificial Analysis measured a one-point score gain over Carbon at the highest tested settings, alongside lower cost and task time; the cost comparison changes while Carbon is discounted through November 30.
Google released Gemini 4 Neon on November 18, pitching it as a balance of depth and efficiency. Independent evaluator Artificial Analysis measured Neon at 61 points on its Intelligence Index at xhigh effort, one point above Gemini 4 Carbon at high, while a standard index task cost less and took less time in its comparison. Whether that makes Neon a better buy depends on workload, settings and Carbon’s temporary discount.
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
On the Artificial Analysis Intelligence Index, Neon at xhigh scored 61, compared with Carbon at high scoring 60. These are the best tested settings shown in the source material, so the one-point difference is not a same-setting comparison. At the same high setting, Artificial Analysis says Neon matched Carbon’s score for $3.74 per task. At the two models’ respective top tested settings, its index task cost $6.44 with Neon and $7.76 with Carbon, and took 21.5 minutes rather than 37.1.
Neon’s standard token rates are $3 per million input tokens and $15 per million output tokens. Carbon’s standard rates are $4 and $20, but Google’s discount lowers Carbon to $2 and $10 through November 30. That makes Carbon cheaper per token during the promotion, even though Artificial Analysis’s stated task comparison at standard prices favors Neon. Actual costs can vary with the workload and token use.
Google’s launch post says Neon brings “frontier-level intelligence with fast end-to-end task completion.” Its comparison table reports Neon ahead in 16 of 19 rows against Claude Opus 5.5 and GPT-6.2 Sol. The source article notes that those are not the two models leading the cited independent Intelligence Index rankings. Separately, on Artificial Analysis’s Coding Agent Index, Neon scored 58.6, close to Claude Fable 5.2 at 58.5 and Claude Opus 5.5 at 58.2; reported task costs were $6.86, $15.47 and $12.61, respectively.

Where Neon Could Cut Costs
The results make Neon a candidate for teams that run repeated, substantial model tasks and pay for both inference and waiting time. In Artificial Analysis’s index task comparison, Neon was $1.32 cheaper and finished 15.6 minutes sooner than Carbon at the listed standard prices and tested settings. The Coding Agent Index results also suggest that Neon’s performance was close to the two Claude models cited while its reported task cost was lower. These are measurements from particular evaluations, not a guarantee of savings on every company’s workload.
The purchasing decision is therefore about more than the list price or a one-point index change. Teams should compare their own prompts, output volume, quality requirements and effort settings, and account for the Carbon promotion through November 30. A small score difference may matter little for routine work, while throughput and cost could matter more for high-volume use.
Products Worth Considering

Apple Watch SE 3 GPS + Cellular 40mm Smartwatch

Bose SoundTouch 30 Wireless Speaker, Black

Bose TV Speaker Compact Soundbar with Bluetooth and HDMI-ARC, Black
As an affiliate, we earn on qualifying purchases.
Four Gemini 4 Releases
Neon is the fourth Gemini 4 model Google has released in roughly seven weeks: Argon on September 30, Carbon on October 28, Hydrogen on November 12 and Neon on November 18. The model card says Neon is adapted from Gemini 4 Carbon; Google directs readers to Carbon’s documentation for architecture, training data and hardware details.
Google describes Carbon as intended for difficult, long-horizon projects and Hydrogen for fast everyday execution, with Neon between them. Its model card lists text, image, audio and video inputs, text output and a one-million-token context window. Neon is available through the Gemini API in AI Studio and several Google products, with access depending on the product and subscription. Google says a cyber-defense version is being made available to eligible trusted defenders through its Fairwind Program.
“frontier-level intelligence with fast end-to-end task completion.”
— Google, in its Gemini 4 Neon launch post
Limits of the Comparisons
The source figures do not establish how Neon performs on every customer workload, nor do they show whether the one-point Intelligence Index difference is meaningful in day-to-day use. The top-setting comparison uses different effort levels for Neon and Carbon; the same-high comparison is described separately. The methodology and workload mix behind each company’s own benchmark table also affect how broadly those results can be applied.
It is also unclear from the supplied material how costs compare for a specific organization once its prompt patterns, cached inputs, output lengths and negotiated prices are included. Carbon’s discount expires November 30 according to the source, but the material does not say whether Google will extend or replace it. No independent results are provided here for the cyber-defense version.
Check Pricing After November
Businesses weighing a switch can compare Neon and Carbon against representative tasks before committing, tracking quality, completion time and total token use at the effort setting they plan to deploy. They should calculate costs using their actual API rates and include any cached-input savings. November 30 is the stated end date for Carbon’s $2 input and $10 output token promotion; the price after that date is not specified in the source material.
Further independent evaluations and Google’s future pricing announcements could change the comparison. For now, the available evidence points to a modest index-score difference, with task cost and speed offering the more substantial measured distinction under the cited test conditions.
Key Questions
How much higher did Neon score than Carbon?
Artificial Analysis scored Neon at 61 at xhigh effort and Carbon at 60 at high effort. At the same high setting, the source says Neon matched Carbon’s score.
Is Neon cheaper than Carbon?
Neon’s standard rates are $3 per million input tokens and $15 per million output tokens, below Carbon’s standard $4 and $20. Carbon is temporarily discounted to $2 and $10 through November 30, so it currently has lower token prices during that offer.
How did Neon perform on coding tasks?
On the Artificial Analysis Coding Agent Index, Neon scored 58.6, compared with 58.5 for Claude Fable 5.2 and 58.2 for Claude Opus 5.5. The cited task costs were $6.86 for Neon, $15.47 for Fable 5.2 and $12.61 for Opus 5.5.
What is still unknown about Neon’s value?
The supplied comparisons do not show how Neon will perform or cost on every business workload. Results depend on task type, effort setting, token use and applicable pricing, and the source does not specify Carbon’s price after its discount ends.
Source: Thorsten Meyer AI
