Inter-office memo

To
Everyone who wants Gemini 4 Neon for less
From
Mo, Chief Executive
Date
November 20, 2026, 4:44 AM CST
Re
The cheapest way to use Gemini 4 Neon

The cheapest way to use Gemini 4 Neon for coding in 2026: 8 ways compared

For most people it’s Google AI Pro, $19.99 a month, where Neon is Antigravity’s default and Orchestrator (Neon) hands routine edits to the cheaper Hydrogen. The Gemini API, $3 per million input tokens and $15 output, suits a few tasks a month, and Pro’s $10 of monthly Cloud credits can pay for some of them; Copilot Pro+, $39, suits people already in Copilot. No free plan has Neon.

Disclosure: Enjoy is our product, and it runs Antigravity on your own Google plan. We checked every plan and price below against each company’s own pages on November 20, 2026, and we say where you don’t need Enjoy at all, which is most of the time.

Gemini 4 Neon compared

Gemini 4 Neon is the Gemini model Google’s Neon guide places “between Gemini 4 Hydrogen and Gemini 4 Carbon for high-frequency complex tasks.” It came out on November 18, 2026, and became the default model for new conversations in Antigravity on paid plans. Our memo covers the launch. In short: Google AI Pro is the cheapest way for most people, the API is cheaper only for a few tasks a month, and Hydrogen does the routine part for less. These are the ways to pay for Neon, side by side.

Way Best for Price, checked November 20, 2026 Where you use Neon How much you get
Google AI Pro Most people $19.99 a month Antigravity’s app, CLI and IDE extensions; the Gemini app; Google AI Studio A quota “refreshed every five hours until the weekly limit is reached”
Pro plus AI credits Running out of Pro now and then $19.99 a month, plus AI credits from Google One Antigravity Pro’s quota, then credits
Google AI Ultra Running out of Pro most weeks, or wanting Carbon too $99.99 or $199.99 a month The same as Pro “The highest, most generous quota, refreshed every five hours”
The Gemini API A few tasks a month, scripts and other agents $3 per million input tokens, $15 per million output; half on Batch and Flex; no monthly fee Antigravity CLI with a key, OpenCode, OpenRouter, your own code What you pay for
Cloud credits with Pro or Ultra Subscribers who also use the API Included: $10 a month with Pro, $40 or $100 with Ultra The Gemini API, with your paid key Up to the credit, then API rates
GitHub Copilot Pro+ People who already work in Copilot $39 a month, with $70 of AI credits Copilot in VS Code and other editors, its CLI, coding agent and app Credits, at $3 and $15 per million tokens
Cursor Premium People who already work in Cursor $30 a month, plus prepaid credit for Neon Cursor Prepaid credit, at $3 and $15 per million tokens
Enjoy, with Google AI Pro Making a Google plan go further, beside a Claude or ChatGPT plan A monthly subscription on top of your plans, free for 7 days (pricing) Antigravity, on your Google plan Your Google plan’s

Prices are each company’s US list prices, before tax. Google doesn’t publish how big any Antigravity quota is or how much of it a Neon turn uses, so those cells are in its own words.

Is there a free way to use Gemini 4 Neon?

No. Every free plan stops short of it:

  • The Gemini API’s free tier doesn’t have it: “Gemini 4 Neon is not available on the Free tier,” says Google’s pricing page.
  • Antigravity without a plan, and with Google AI Plus, doesn’t either. Google’s models page says “Gemini 4 Neon isn’t available on Free and Plus plans,” and Plus’s own line on Google’s plans page is “Access to Gemini 4 Hydrogen.”
  • A Google AI Pro trial gets it only sometimes. Neon is “Available with paid Google AI Pro plans or on trial with eligible premium devices,” and Google’s Antigravity pages don’t say which devices count. Google also offers students “a free Pro plan for one year”; its Antigravity pages don’t say whether that counts as paid or as a trial.
  • GitHub Copilot Free and Pro aren’t among the plans GitHub’s changelog names for Neon.
  • Cursor Free and Plus don’t include it: “Plus and Free don’t include third-party models,” says Cursor’s models page.
  • Kiro lists no Gemini 4 model on its models page.

The nearest free thing is Hydrogen. On the API’s free tier it’s “Free of charge,” input and output, though Google says free-tier content is “used to improve our products.” In Antigravity, though, Hydrogen needs a paid plan too.

What Neon costs, and when Hydrogen will do

At Google’s API prices, Neon is $3 per million input tokens, $0.15 per million cached and $15 per million output, and the output price includes thinking. It “uses flat pricing across the full 1M context window,” and Google’s pages give no end date for that price, unlike Argon and Carbon, which go from $2 and $10 to $4 and $20 on December 1. Batch and Flex requests cost half: $1.50 and $7.50. Hydrogen is $1.50 and $7.50 at the standard rate, half of Neon.

Coding agents read and write a lot. Artificial Analysis’s leaderboard puts Neon at xhigh, its top thinking level, at $6.44 and 21.5 minutes a task on its Intelligence Index, against $7.76 and 39.9 minutes for Carbon at high. At that rate, four tasks cost more than a month of Google AI Pro. Google’s own announcement says Neon at its default high level has “a 32% lower cost per task than Carbon.”

Hydrogen is the bigger saving for routine work, and in Antigravity you don’t have to choose by hand. Orchestrator (Neon) “uses Neon for upfront planning, architecture, and complex reasoning while handing off routine code edits, package updates, and test runs to Hydrogen,” says Google’s Antigravity post, “stretching your quota further on every run.” It’s on paid Pro and Ultra, and it “runs with a fixed configuration, so you can’t choose a thinking level for it.”

So Pro is the cheapest Neon if you code most days, Orchestrator (Neon) stretches it furthest, and the API wins only if you use Neon a few times a month.

Google AI Pro

Best for: most people who want Gemini 4 Neon for coding.

Pro is the cheapest plan with Neon. In Antigravity, “It’s the default model for new conversations if you haven’t chosen a model; if you’ve already selected one, Antigravity keeps your selection,” with “Up to 512K tokens” of context, half the API’s. Google One’s plan page adds “Expanded access to Gemini 4 Neon in Google Antigravity and Google AI Studio” and “Access to Gemini 4 Neon for deeper analysis in the Gemini app,” where Google’s limits page gives Pro “4x higher than standard limits.” Antigravity’s plans page gives Pro “A high, generous quota, refreshed every five hours until the weekly limit is reached.” Our Pro vs Ultra guide covers the rest of the plan.

Price: $19.99 a month.

Watch for: Google publishes no quota sizes. The limits “correlate with the amount of work done by the agent,” so “more complex tasks consume more quota,” and “Usage limits for this service are subject to modification.” A trial gets Neon only “with eligible premium devices”; to switch to a paid plan, Google says to “submit a request.” And on November 19, Google AI Ultra subscribers reported Neon’s Extra level failing with “Thinking level is unsupported: THINKING_LEVEL_XHIGH” on Google’s forum. One of them wrote later that day that “it is working again for me”; Google hadn’t replied in either thread.

Pro plus AI credits

Best for: people who run out of Pro now and then, when Ultra would sit unused most weeks.

“Users on Google AI Pro or Ultra plans can use purchased AI credits (or any one-time promotional credits) for additional overage usage above the baseline quota,” says Antigravity’s plans page. The AI Credit Overages setting decides when: Never waits “until the baseline quota refreshes,” and Always uses credits “when the baseline quota is exhausted (automatically switches back to using the baseline quota once it refreshes).” In Antigravity CLI, it’s the Use AI Credits field under /config, says its credits page. You buy credits from Google One, in amounts such as “2,500, 5,000, or 20,000 AI credits.”

Price: Pro, plus the credits you buy. “AI credits are consumed at standard Gemini Enterprise consumption pricing,” says the plans page; none of the pages we read gives a credit’s price in dollars or how many a Neon turn uses.

Watch for: not everyone can buy them. “AI credits are not available for purchase for Google AI Plus plans” or “in Japan,” and Google One members “subscribed through a third party, Pixel Pass, or Apple billing” can’t either. “AI credits may expire after a certain period, as set out when you acquire them.” With Always on, Antigravity moves to credits without stopping.

Google AI Ultra

Best for: running out of Pro’s weekly limit most weeks, or wanting Gemini 4 Carbon beside Neon.

Ultra has “Higher access to Gemini 4 Neon in Google Antigravity and Google AI Studio” and “The highest, most generous quota, refreshed every five hours,” with “The highest weekly rate limits.” It’s also the plan with Carbon, which Antigravity’s pricing page lists beside Neon as Ultra’s “flagship” models. It comes in two sizes, sold on Google’s plans page with “5x higher usage limits vs. AI Pro” and “20x higher usage limits vs. AI Pro,” multiples the Gemini app’s limits page applies to the app.

Price: $99.99 a month, or $199.99 for the larger size.

Watch for: Google’s May post on Antigravity’s plans puts Ultra at “5X worth of tokens” and “20X worth of tokens” compared to Pro, so Ultra 5x costs the same per token as Pro and only buys more of it. Try credits before you move up for Neon alone.

The Gemini API

Best for: a few Neon tasks a month, scripts, and agents that take an API key.

The API has no monthly fee, and paid-tier content is “not used to improve our products.” Batch requests and Flex requests, an “inference tier that offers a 50% cost reduction compared to standard rates,” halve the price for work that can wait. Repeated input is discounted too: “Implicit caching runs by default on prompts of at least 4,096 tokens and discounts cached input tokens by 95% on cache hits,” says Google’s Neon guide.

You can use a key in several places. Google’s install guide says you can “Run Antigravity CLI with your own Gemini API key instead of a signed-in Google account,” and that “Model requests go directly to the Gemini API.” For other agents, Google’s FAQ says “To use a third-party coding agent with Gemini, we recommend using a Gemini Enterprise or Google AI Studio API key.” “OpenCode lists models from models.dev,” says its models page, and models.dev lists gemini-4-neon under its Google provider at $3 and $15. OpenRouter lists Google AI Studio and Vertex among its routes for Neon, at $3 and $15.

Price: $3 per million input tokens, $0.15 cached and $15 per million output; $1.50 and $7.50 on Batch and Flex.

Watch for: thinking is billed as output, at $15 per million, and a coding agent runs many turns for one task. Flex “is in preview,” with a target of “Minutes (1-15 min target)” and “Best-effort (Sheddable)” reliability, so it suits scripts, not a conversation. In Antigravity CLI, “Only setting a GEMINI_API_KEY environment variable on its own has no effect”: you set "modelProvider": "gemini" too, and the install guide doesn’t list which models work that way. And signing other tools in with your Antigravity login isn’t the same thing: Google’s FAQ says it “can result in suspension or termination of your account.”

Cloud credits with Pro or Ultra

Best for: Pro and Ultra subscribers who also call Neon through the API.

Pro and Ultra come with monthly Google Cloud credits, and since November 13 you claim them in Google AI Studio: “$10 every month” on Pro, and “Ultra 5x receive $40 and Ultra 20x earns $100 every month,” says Google’s claim page. “Your credits cover Gemini API calls with your paid key, for models and agents in Playground or for Gemini API powered AI features in your own apps.” Each grant “is valid for 12 months from issue,” and “Grants stack month to month,” says Google’s announcement. Our memo shows how to claim them.

Price: included in Pro and Ultra; “Google AI Plus does not include credits from the Google Developer Program.” You first need “a $5 minimum prepayment and balance in order to draw down promotional credits.”

Watch for: they’re small next to Neon’s prices: Pro’s $10 covers about one and a half of Artificial Analysis’s $6.44 tasks. Google’s pages describe AI Studio and your own apps, and don’t mention coding agents. They don’t add Antigravity quota: the Google One credits that pay for overage are “a separate credit system that isn’t supported in AI Studio,” says Google’s credits guide. And “Only the primary plan subscriber can use these credits. They can’t be shared among family members.”

GitHub Copilot Pro+

Best for: people who already work in Copilot and want Neon there.

“Gemini 4 Neon is available to Copilot Pro+, Max, Business, and Enterprise users,” says GitHub’s changelog, in VS Code, Visual Studio, Copilot CLI, the coding agent, the GitHub Copilot app, github.com, GitHub Mobile, JetBrains IDEs, Xcode and Eclipse. “This model is billed at provider list pricing under usage-based billing”: $3 per million input tokens, $0.15 cached and $15 per million output, in AI credits, where “1 AI credit = $0.01 USD,” per GitHub’s price table.

Price: Pro+ is $39 a month, with “$70 monthly total credits”: $39 of base credits and a $31 flex allotment, says its plans page. Max is $100, with $200.

Watch for: the flex part isn’t fixed: “Flex allotments may change over time.” “Rollout will be gradual,” and on November 20 the plans page’s model lists didn’t show Neon yet. Copilot bills Neon at the API’s prices, so it’s no cheaper per token, only simpler if you already pay for Pro+. Our Copilot plan guide compares the plans.

Cursor Premium

Best for: people who already work in Cursor and want Neon now and then.

“Gemini 4 Neon is a third-party model, so it isn’t part of the usage included with Cursor plans,” says Cursor’s Neon page. On Premium and above, its models page says, third-party models are “available in Cursor at the model’s API list price from the first request, paid from your prepaid credit”: for Neon, $3, $0.15 and $15 per million tokens.

Price: Premium is $30 a month, or $300 a year, plus the prepaid credit you spend on Neon.

Watch for: you pay the API’s price for Neon on top of the plan. Cursor’s pricing page says “Premium and above include generous usage of Grok models but do not include an allowance for third-party models,” and “Plus includes Grok models only.” If Neon is the reason you’d buy Cursor, Google AI Pro costs less.

Enjoy, with Google AI Pro

Best for: making a Google plan go further, with a Claude or ChatGPT plan beside it.

Enjoy is a desktop app for Mac, Windows and Linux that runs Antigravity on your own Google account, beside Claude Code, Codex, Grok Build, Muse Code and local models through OpenCode, each on its own lab’s plan. Its Antigravity model list shows “the Gemini models Google offers your account,” and its Antigravity guide lists Neon’s Extra level as Extra high. It doesn’t sell Neon or make it cheaper per token. What it does for the bill:

  • Shows what’s left. The agent settings show the limits Google shows for each group of models, such as Gemini · Weekly, with when each resets.
  • Waits instead of paying. When Antigravity hits a limit, Enjoy waits until the limit resets, then tries again, and the agent continues where it left off, so work you leave running doesn’t need credits.
  • Hands work to another lab. Switch the conversation to Claude Code or Codex on a plan you already have, and “The new agent gets the conversation so far, including replies from other agents.”
  • Uses an API key where you want one. OpenCode in Enjoy uses the providers you set up in it, so it can run Neon on your own Gemini API key, at Google’s prices.

Price: Enjoy is one monthly subscription on top of your AI plans, and the first seven days are free, with a card to start. It includes your phone, any browser and one computer; each other computer costs extra each month. See pricing.

Watch for: it’s another bill, so it pays only if it saves you credits, an upgrade or time. Enjoy doesn’t add to Google’s quota, and it can’t give Neon to a Free or Google AI Plus account. Antigravity uses one Google account in every project, so Enjoy can’t move it to a second account the way it does for Claude Code, Codex and Grok Build. OpenCode shows Usage unavailable, so check what a key spends with Google. Enjoy isn’t open source, and a waiting conversation resumes only once your computer is awake with Enjoy open.

When to stay on your current plan

  • On Google AI Pro, and rarely running out: stay. It’s Google’s cheapest plan with Neon, and Orchestrator (Neon) makes the quota go further.
  • On Ultra for Carbon: stay. Neon is included, and you’d lose Carbon by moving down.
  • In Copilot Pro+ or Cursor Premium every day: if you reach for Neon only now and then, their credits are simpler than a second subscription.
  • Calling Neon only from scripts: stay on the API, and use Batch or Flex for anything that can wait.

How we compared

We read Google’s Neon announcement and Antigravity post; Antigravity’s models, plans, pricing, FAQ, install, CLI credits and changelog pages; Google’s AI plan pages, Gemini app help on limits and models, and Google One’s help on AI credits; the Gemini API’s pricing, Neon model, Neon guide and Flex pages, and Google’s pages on Cloud credits in AI Studio; GitHub’s changelog, Copilot plans and price table; Cursor’s Neon, models and pricing pages; Kiro’s models page; OpenRouter’s Neon page, the models.dev catalog and OpenCode’s providers and models pages; Artificial Analysis’s leaderboard; and two threads on Google’s developer forum, all on November 20, 2026. We quote each company’s own words for its limits, and we don’t estimate what a company hasn’t published, such as how big a quota is or what an AI credit costs. Prices change often, so we recheck this guide against its sources at least once a month, and the date at the foot of the page says when we last did.

14.Questions.

Is there a free way to use Gemini 4 Neon?

No. Google’s API pricing page says “Gemini 4 Neon is not available on the Free tier,” and Antigravity and the Gemini app offer it on Google AI Pro and Ultra, not on Free or Google AI Plus. A Pro trial gets it only “with eligible premium devices,” and GitHub Copilot’s and Cursor’s free plans don’t include it.

What’s the cheapest subscription with Gemini 4 Neon?

Google AI Pro, at $19.99 a month, with Neon as the default model in Antigravity, in the Gemini app and in Google AI Studio. Cursor Premium, at $30, and GitHub Copilot Pro+, at $39, have it too, but both bill Neon at Google’s API prices.

Is the Gemini API cheaper than Google AI Pro for Neon?

Only for light use. Neon costs $3 per million input tokens and $15 per million output, thinking included, and Artificial Analysis puts a task on its Intelligence Index at $6.44 with Neon at xhigh, so four such tasks cost more than a month of Pro. Batch and Flex requests cost half, if the work can wait.

Should I use Neon or Hydrogen to save money?

Hydrogen, for routine work. On the API it costs half what Neon does per token and is free on the free tier. In Antigravity, Orchestrator (Neon) does this for you, using Neon “for upfront planning, architecture, and complex reasoning while handing off routine code edits, package updates, and test runs to Hydrogen.”

Do Google AI Pro’s Cloud credits pay for Gemini 4 Neon?

On the Gemini API, yes. Pro includes $10 of Google Cloud credits a month and Ultra $40 or $100, which “cover Gemini API calls with your paid key,” after a $5 prepayment. They don’t add Antigravity quota, which runs on Google One’s separate AI credits.

Can I use Gemini 4 Neon in GitHub Copilot or Cursor?

Yes, at Google’s API prices. GitHub made it available to Copilot Pro+, Max, Business and Enterprise on November 18, billed at $3 and $15 per million tokens in AI credits. Cursor offers it on Premium and above from prepaid credit, and says “Plus and Free don’t include third-party models.”

Enjoy Agentic Industries · Checked against its sources on