🔍 Read the full analysis: The Gemini 4 Neon Table, Decoded: Gains, Savings, And Missing Details 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, 2026, positioning it as a balance of capability and efficiency. Artificial Analysis measured a one-point score gain over Gemini 4 Carbon at the highest settings it compared, alongside lower cost and shorter task times; the results vary by effort setting and benchmark. Google’s published comparison omits two models that lead the independent index, and several details about Neon’s architecture remain undisclosed.
Google released Gemini 4 Neon on November 18, describing it as a model that balances capability and efficiency; independent testing found a one-point score gain over Gemini 4 Carbon at the compared highest settings, as well as lower cost and shorter completion times. The results are not a broad win across every test, and Google’s published benchmark table leaves out two models that rank ahead of Neon on the independent index.
Artificial Analysis gives Neon a score of 61 at xhigh effort, compared with Carbon’s 60 at high effort. On its index tasks, Neon took an average of 21.5 minutes and cost $6.44, versus 37.1 minutes and $7.76 for Carbon at its standard list price. Because the models were measured at different effort settings in that comparison, the figures do not isolate a model-only difference. At the same high setting, Neon scores 60 and costs $3.74 per task, according to Artificial Analysis.
Neon’s list price is $3 per million input tokens and $15 per million output tokens, below Carbon’s standard $4 and $20 rates. Carbon is discounted to $2 and $10 through November 30, according to the source material, so Neon is not cheaper than that temporary offer on those token prices. Artificial Analysis also reports that Neon generates output at about 110 tokens per second, compared with about 55 for Carbon, while using roughly 15% more output tokens per task. Those measurements help explain why the price difference does not translate directly into the same percentage reduction in task cost.
The model’s effort setting changes the measured trade-off. Artificial Analysis reports scores and costs of 61 and $6.44 at xhigh, 60 and $3.74 at high, 55 and $1.90 at medium, and 44 and $0.75 at low; reported task times range from 21.5 minutes at xhigh to 3.7 minutes at low. These are index-specific results at list prices, not a promise of cost or speed for every user workload. Google says Neon accepts text, images, audio and video, has a one-million-token context window, and outputs text, with a March 2026 knowledge cutoff.
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
Neon’s Cost-Speed Trade-Off
For businesses choosing a model for repeated coding or other multi-step work, cost and elapsed time per task may matter as much as a small change in a benchmark score. Artificial Analysis places Neon close to other models on its coding-agent index: Neon scores 58.6, compared with 58.5 for Claude Fable 5.2 and 58.2 for Claude Opus 5.5. Its reported coding-agent task cost is $6.86, versus $15.47 for Fable 5.2 and $12.61 for Opus 5.5. Those figures offer a reason for teams to test Neon, but they are index results, not guaranteed savings in production.
The range across effort settings also matters for deployment decisions. On the reported index, moving from high to xhigh raises the score by one point and the task cost from $3.74 to $6.44. A buyer may prefer the lower setting for routine jobs and reserve higher effort for harder tasks, but the source material does not establish which setting performs best on any particular company’s work. Real costs can depend on prompts, output length, caching, and usage patterns.
Products Worth Considering

Apple MacBook Air 13-inch with M5 Chip, 16GB Memory, 512GB SSD, Liquid Retina Display, and 12MP Center Stage Camera – Midnight

DJI Mini 4K Drone Combo with 4K Camera, 3-Axis Gimbal, and Two Batteries

Kindle Paperwhite Signature Edition 32GB (2024 Model) – Metallic Black
As an affiliate, we earn on qualifying purchases.
Google’s Rapid Gemini 4 Releases
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 aimed at fast everyday execution, with Neon between them. Its model card says Neon is adapted from Gemini 4 Carbon; for architecture, training data and hardware, it points readers to Carbon’s documentation rather than supplying new details.
Google’s evaluation document reports Neon as best in 16 of 19 benchmark rows against Claude Opus 5.5 and GPT-6.2 Sol. In the three rows where a competitor leads, the document shows Opus ahead on Terminal-bench 4.1 and FrontierCode 2.0, and GPT-6.2 Sol ahead on Chartography. The comparison is limited to those named competitors. Artificial Analysis ranks GPT-6.2 Astra and Claude Fable 5.2 ahead of Neon on its overall index, but Google’s table does not include them.
Independent results also show that Neon does not lead every test against Carbon. On Terminal-Bench 4.0, Neon at xhigh scored 63.6%, below Carbon at high with 65.2%. On Artificial Analysis’s AA-Omniscience test, Neon had a reported hallucination rate of 14.4%, compared with Carbon’s 11.4%. Google separately claimed Neon had its lowest fabrication rate yet in internal Antigravity testing; that is a different measure and the source says outside readers cannot verify it.
“frontier-level intelligence with fast end-to-end task completion”
— Google, in its launch post
Benchmark Gaps and Model Details
The available comparisons do not settle how Neon performs across all workloads. Google’s table uses Claude Opus 5.5 and GPT-6.2 Sol, rather than the two models ahead of Neon on Artificial Analysis’s index, and benchmark scores can depend on evaluation methods and settings. The source material does not provide a direct, fully matched comparison across all leading models at identical effort levels, prices, and test conditions.
Further details about Neon’s architecture, training data and hardware are also not supplied in its model card, which refers readers to Carbon’s documentation. The reported internal fabrication result from Google has no outside verification in the cited material. It is also not clear how closely index costs and speeds will match the experience of customers using their own prompts, tools and output lengths.
Availability and Buyer Testing
Google lists Neon in the Gemini API through AI Studio, Gemini for business, and Google Antigravity for paid AI Pro and Ultra subscribers. It is also available in the Gemini app to Pro and Ultra subscribers. Google says Gemini 4 Neon for cyber will reach eligible trusted defenders in its Fairwind Program over the next few days, as of the November 18 release.
The next useful evidence will come from matched evaluations and customer workloads that show performance, reliability, and cost at the effort setting users actually choose. Buyers comparing list prices should also account for Carbon’s stated discount through November 30. No further benchmark release or timeline for additional model documentation is specified in the source material.
Key Questions
What did Google announce?
Google released Gemini 4 Neon on November 18, 2026. It describes Neon as a model between Carbon and Hydrogen in its Gemini 4 lineup; the model card says Neon is adapted from Carbon.
Does Neon outperform Gemini 4 Carbon?
Not on every measure. Artificial Analysis reports a score of 61 for Neon at xhigh and 60 for Carbon at high, but Neon scores lower on Terminal-Bench 4.0 in the cited comparison. At the same high setting, Neon scores 60 on the index.
Is Neon cheaper than Carbon?
Neon’s standard token prices are lower than Carbon’s standard rates: $3/$15 per million input/output tokens, versus $4/$20. Carbon is discounted to $2/$10 through November 30 according to the source, so that temporary offer is lower. Artificial Analysis reports a $3.74 Neon index task at high effort, but task costs depend on the evaluation and usage.
Which leading models are missing from Google’s comparison table?
Google’s 19-row table compares Neon with Claude Opus 5.5 and GPT-6.2 Sol. It does not include Claude Fable 5.2 or GPT-6.2 Astra, which rank ahead of Neon on the Artificial Analysis Intelligence Index.
Where can people use Gemini 4 Neon?
Google lists Neon in the Gemini API through AI Studio, Gemini for business, Google Antigravity for paid AI Pro and Ultra subscribers, and the Gemini app for Pro and Ultra subscribers. A cyber-defense version is planned for eligible members of Google’s Fairwind Program.
Source: ThorstenMeyerAI.com
