Gemini 4 Neon: Google’s $3 Model Takes on Anthropic

Google launched Gemini 4 Neon on 18 November, priced at $3 per million input tokens.

It is the fourth Gemini 4 model in under two months, and it sits between the heavy Carbon model and the fast Hydrogen model. On the hardware side, Nvidia has priced its AI laptops for India: Nvidia RTX Spark Laptops in India From Rs 2.99 Lakh.

Updated: 20 November 2026, 10:45 IST

Quick Read

  • Gemini 4 Neon costs $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window.
  • It is adapted from Gemini 4 Carbon, but built for fast, high-frequency work such as coding and knowledge tasks.
  • Neon is live in the Gemini API, Google AI Studio, and the Gemini app for Pro and Ultra subscribers. It is not on the free tier.

Google launched Gemini 4 Neon on 18 November 2026, per Google’s own blog. It costs $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window. It is the fourth Gemini 4 model in under two months.

One note on the date: Google’s blog is dated 18 November, and most outlets agree. Geeky Gadgets dates the release 19 November, which looks like a time-zone difference. This report uses 18 November, the date on Google’s post.

Where Neon sits in the family

Google now has four Gemini 4 models. Argon came first, on 13 October. Carbon followed on 10 November as the strongest. Hydrogen arrived on 12 November as the fast, cheap one. Neon, on 18 November, is the fourth.

Neon sits in the middle. Google built it from Carbon, so it keeps much of Carbon’s power, but tuned for speed: complex coding, knowledge work, and science jobs that run again and again.

Google says Neon brings “frontier-level intelligence with fast end-to-end task completion”. In plain words, it takes on hard jobs and finishes them faster than Carbon would. Google also says Neon can still hallucinate, inventing facts that sound right but are wrong, so users should check answers that count.

Neon has four thinking levels, per the model card, so a developer can pick how hard it thinks. Thinking tokens are billed as output.

Also read: Anthropic IPO: $81 billion, biggest ever

The price has a twist

Neon costs $3 per million input tokens and $15 per million output tokens. Cached input, text the model has already seen, costs $0.15 per million tokens, a 95% discount, per Google.

Here is the twist: until 30 November, Neon is not the cheap option. Carbon and Argon cost $2 and $10 on their launch offer, which runs through 30 November. So until that date, Neon costs 50% more than Carbon per token.

From 1 December, Carbon rises to $4 and $20, while Neon stays at $3 and $15, 25% below Carbon’s full price. Google says Neon also costs about 32% less per task than Carbon at its default setting, because it finishes jobs with fewer tokens.

Hydrogen is still the cheapest: $1.50 and $7.50, half of Neon’s price, trading a little quality for speed. Neon is for jobs where Hydrogen is not quite strong enough but Carbon is more than the job needs.

Who can use it, and where

Neon is rolling out in the Gemini API in Google AI Studio, in Gemini for business, and for paid Pro and Ultra subscribers in Google Antigravity. Pro and Ultra subscribers also get it in the Gemini app.

It is not on the free tier: Google’s pricing page says Neon is “not available on the Free tier”, per FourWeekMBA. Free-tier users get Hydrogen instead, through auto-selection.

The rollout is global, and that includes India: Pro and Ultra subscribers here get Neon at the same time as users elsewhere. Google has named no India-only price, so the dollar token prices above apply.

How it scores

Early test scores put Neon near the top. On the Artificial Analysis Intelligence Index, Neon scores 61 at extra-high thinking, per VentureBeat’s table from Google’s data — one point above Carbon’s 60 and well above Hydrogen’s 54, and first on Vending Bench 2 and Zapier’s Automation Bench.

Google also links to coding tests that put Neon one point above Anthropic’s Claude Opus 5.5 and GPT-6.2 Sol on extra-high reasoning. Those are Google’s own linked tests, not third-party checks. Geeky Gadgets notes it has not run deep coding tasks on Neon to check the claim.

Hydrogen is still the quick one: Google says it runs 2.5 times faster than Gemini 3.8 Flash, with a 1,048,576-token window. Carbon, launched 10 November, is the strongest model, built for cybersecurity work and long jobs where an AI agent works on its own for hours. Neon borrows Carbon’s strengths and trims the cost and wait.

Also read: IBM’s $3 billion India quantum and AI jobs plan

The timing is no accident

Neon landed one day before Anthropic’s shares started trading, and Anthropic’s Claude Opus 5.5 is the model Google’s Neon tests are measured against. Four models in 36 days is a fast pace. Google is splitting its line into clear jobs: Hydrogen for speed, Carbon for the hardest work, Neon for the middle ground where most paid work sits.

Desk-original added context

The table below is worked out by this desk from Google’s published prices. It shows what each model costs for the same job, and how prices change on 1 December.

ModelInput per 1M tokensOutput per 1M tokensCost for 1M in + 1M out
Hydrogen$1.50$7.50$9.00
Argon / Carbon (to 30 Nov)$2.00$10.00$12.00
Neon (flat)$3.00$15.00$18.00
Carbon (from 1 Dec)$4.00$20.00$24.00

Take a job using 1 million input and 1 million output tokens. On Hydrogen it costs $9. On Neon $18, twice as much. On Carbon after 1 December, $24. So Neon sits exactly in the middle: twice Hydrogen, 25% below Carbon’s full price.

But timing matters: until 30 November, Carbon costs $12 for the same job, less than Neon’s $18. A developer who picks Neon this week pays 50% more per token than one on Carbon’s launch price. From 1 December, Neon is the cheaper of the two.

Cached input on Neon costs $0.15 per million tokens, 95% off the $3 input price. A job that reuses the same long document, such as a code base, will pay far less than one that feeds in fresh text each time.

FAQ

When did Google launch Gemini 4 Neon?
Google launched Gemini 4 Neon on 18 November 2026, per its own blog post. Geeky Gadgets dates it 19 November, which looks like a time-zone difference. This report uses 18 November, the date on Google’s post.

How much does Gemini 4 Neon cost?
Neon costs $3 per million input tokens and $15 per million output tokens, with cached input at $0.15. From 1 December that is 25% below Carbon’s full $4 and $20, but until 30 November Carbon’s $2 and $10 launch price is cheaper.

Can free users in India use Gemini 4 Neon?
No. Google’s pricing page says Neon is not available on the free tier. Free-tier users in India get Hydrogen through auto-selection. Neon needs a Pro or Ultra subscription, or access through the paid API.

Also read: Bengaluru Tech Summit 2026: AI and beyond

Sources

3 thoughts on “Gemini 4 Neon: Google’s $3 Model Takes on Anthropic”

  1. Pingback: Anthropic $81.3 Billion IPO Is Biggest Ever

  2. Pingback: Apple Builds AI Agent to Rival ChatGPT, Claude

  3. Pingback: Nvidia RTX Spark Laptops in India From Rs 2.99 Lakh

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top