Google launched Gemini 4 Neon on 18 November 2026. Its launch post says the model is priced at $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window.
Our number: that is 25% below the $4 and $20 that Google’s Gemini API pricing page lists for Gemini 4 Carbon starting December 1, and 50% above Carbon’s $2 and $10 through November 30, by our arithmetic. Neon costs twice Gemini 4 Hydrogen’s $1.50 and $7.50.
Business Pill · The Model and the Harness
An evergreen Business Engineer explainer on the difference between a model and the harness around it. It teaches the general idea only and does not discuss Google’s launch.
The key insight: Google launched Gemini 4 Neon on 18 November 2026 at $3/$15 per million tokens, 25% below Gemini 4 Carbon’s list price from December 1 and 50% above Carbon’s price through November 30, by our arithmetic.
What Google Launched
Google describes Neon as “bringing frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows”. The post is signed by Koray Kavukcuoglu, SVP at Google DeepMind and Google’s Chief AI Architect.
Google says Neon is “rolling out today” in the Gemini API in Google AI Studio, in Gemini for business, for paid Google AI Pro and Ultra subscribers in Google Antigravity, and for Pro and Ultra subscribers in the Gemini app. The pricing page says Neon is “not available on the Free tier”. It is Google’s second Gemini move this week, after Gemini Tasks went free in six countries.
Cached input costs $0.15 per million tokens, which Google calls “a 95% discount”. Batch and Flex requests are listed at $1.50 and $7.50. We first reported Neon on 17 November, when it appeared in Gemini web code.

What Was Posted on X
Google DeepMind CEO Demis Hassabis announced the launch of Gemini 4 Neon on 18 November 2026, describing frontier-level intelligence and fast task completion and saying it is available to Pro and Ultra users now.
Omar Sanseviero of Google DeepMind posted “Shipping will continue” on 19 November 2026 with a short video of a periodic table that ends on the words “Gemini 4 will return”. The Update box above describes what the video shows.
TestingCatalog quoted that post and read it as a tease that “Gemini 4 Helium (Flash Lite)” arrives next. That label and timing are TestingCatalog’s reading; the post and video do not name a Helium model.
Update, 19 November 2026, 00:44 UTC: Google lists YouTube Premium Family with AI Ultra. Google’s AI plans page says that with a Google AI Ultra subscription “you’ll also get a YouTube Premium Family plan for no additional cost”, covering “you and up to five household members”. A Google AI Pro subscription “comes with YouTube Premium Lite”, which shows fewer ads on most videos. Seen from the UK, the page lists AI Pro at £18.99 a month and AI Ultra “From £79.99/mo”. An X post by @alexmichael says the bundle is new and that Ultra “starts at $99.99/mo” and YouTube Premium Family costs $26.99 a month on its own; Google’s pages do not date the change, and we have not checked those dollar figures, which the post does not tie to a country. Ultra subscribers get Neon in the Gemini app and Antigravity, as above.
Update, 19 November 2026, 03:38 UTC: Artificial Analysis publishes its own Neon numbers. Artificial Analysis’s model leaderboard lists Gemini 4 Neon (Xhigh) at 61 on its Intelligence Index, behind Claude Fable 5.2 (65) and GPT-6.2 Astra (62) and ahead of Claude Opus 5.5 (59). Its cost to run one Index task is $6.44 for Neon, against $26.54 for Fable 5.2, $10.04 for Opus 5.5 and $4.00 for Astra. By our arithmetic, Neon costs about 24% of Fable 5.2 per task and about 64% of Opus 5.5, but about 1.6 times Astra. Its Neon page adds an output speed of 110 tokens a second and says Neon generated 283M tokens on the Index, “very verbose in comparison to the median of 65.1M”. Google’s own benchmark claims are quoted above.
What Google Claims on Benchmarks
Google says Neon ranks “#3 overall with a score of 61 on the Artificial Analysis Intelligence Index” at its xhigh thinking level, and #2 on the Vals Index at 70.9%. It says Neon ranks “#1 overall on the Artificial Analysis Coding Agent Index (58.6)”.
On cost, Google says that at its default high thinking level Neon pairs a high Intelligence Index score “with a 32% lower cost per task than Carbon”. These are Google’s own figures and third-party scores as Google reports them; we did not run the benchmarks.
Google also says a coding benchmark score “more than triples” when Neon runs inside its own Antigravity harness: SWE-sweep goes from 12.8% on the mini-SWE harness to 39.0% in Antigravity.
Update, 19 November 2026, 10:12 UTC: a Google DeepMind staffer hints at more Gemini 4 models. Omar Sanseviero, who builds Gemini, Gemma and AI Studio at Google DeepMind according to Sanseviero’s X profile, posted “Shipping will continue” at 07:51 UTC with a ten-second video.
The video shows a periodic table. Hydrogen, Carbon, Argon and Neon light up first, then Helium and more than a dozen other elements, and it ends on the words “Gemini 4 will return”.
An X post by TestingCatalog read it as “Gemini 4 Helium (Flash Lite) is teased to arrive next”. The post and video do not name a Helium model, and Google’s Gemini API pricing page had no Helium entry, price or date when we checked at 10:10 UTC. The Flash Lite label and the order are TestingCatalog’s reading.
That page lists four Gemini 4 model IDs: Hydrogen, Neon, Carbon and Argon. Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens, and it is the only one of the four with a free tier.
Neon costs $3 and $15. Carbon and Argon cost $2 and $10 through November 30, then $4 and $20 from December 1.
From December 1, output prices across those four will run from $7.50 to $20 per million tokens, a spread of about 2.7 times (our arithmetic).
How the Price Fits Google’s Lineup
Carbon launched at $2 and $10 with its price set to double on December 1. Google’s pricing page still shows that schedule, and it shows the same $2/$10 rising to $4/$20 for Gemini 4 Argon.
Hydrogen sits at $1.50 and $7.50. Neon’s $3/$15 is in between: from December 1 it is the cheaper of the two frontier tiers per token, while Google says it costs less per task as well.
The Structural Read
As we read the pricing page, Google is filling the gap its own December price rise opens: Carbon and Argon go to $4/$20, and Neon arrives below them at $3/$15.
As we read the launch post, Google is selling Neon on cost per task as much as on price per token, and its own harness numbers carry part of that claim.
Google, Introducing Gemini 4 Neon
“The model is priced at $3 per million input tokens ($0.15 per million cached input tokens, a 95% discount) and $15 per million output tokens, with a 1 million-token context window.”
Three Implications
FOR DEVELOPERS Google lists Neon at $3/$15 per million tokens on the paid tier only; the pricing page says thinking tokens are billed as output.
FOR CARBON USERS Google’s pricing page lists Carbon at $2/$10 through November 30 and $4/$20 from December 1.
FOR SUBSCRIBERS Google says Neon is rolling out for Pro and Ultra subscribers in the Gemini app and for paid AI Pro and Ultra subscribers in Antigravity.
The Business Engineer Lens
This story maps onto the Business Engineer framework The Agentic Harness War.
The analysis puts it this way: “Foundation models become relatively commoditized as orchestration layers capture differentiation.”
As we read Google’s post, its own example fits: Neon’s SWE-sweep score goes from 12.8% to 39.0% when it runs inside Google’s Antigravity harness, by Google’s figures. The analysis predates this launch and does not discuss it.
What Is Not Established
Google gives no usage or revenue figures for Neon. The 25% and 50% comparisons are per-token list prices from Google’s page; actual bills depend on how many thinking tokens each model uses, which the pricing page says are billed as output. A cyber version of Neon is “rolling out to trusted defenders in the coming days”, Google says.
The Bottom Line
Google launched Gemini 4 Neon on 18 November 2026 at $3 per million input tokens and $15 per million output tokens: 25% below Gemini 4 Carbon’s list price from December 1 and 50% above Carbon’s price through November 30, by our arithmetic.
97,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. The launch, availability, pricing, benchmark rankings and cost-per-task claims are from Google’s blog post introducing Gemini 4 Neon, read in full on 18 November 2026. The Carbon, Argon and Hydrogen prices are from Google’s Gemini API pricing page, read the same evening. The 25%, 50% and 2x comparisons are our arithmetic on those list prices. Benchmark results are as Google reports them; we did not run them and did not contact Google.
Sources: Google: Introducing Gemini 4 Neon · Google Gemini API pricing









