Inter-office memo

To
Everyone choosing between Google’s and Anthropic’s big plans for coding
From
Mo, Chief Executive
Date
November 20, 2026, 10:49 AM CST
Re
Google AI Ultra vs Claude Max

Google AI Ultra vs Claude Max in 2026: which $100 or $200 plan to buy for coding

Pick Claude Max, $100, or Claude Ultra, $200, for Claude Code with Fable 5.2, and no five-hour limit on verified Ultra. Pick Google AI Ultra, $99.99 or $199.99, for Antigravity with Gemini 4 Carbon and Claude Opus 5.5, 20 or 30 TB of storage, and YouTube Family. Or pay $39.99 for both $20 plans, since paid Google AI Pro has Neon, with Enjoy to hand work across.

Disclosure: Enjoy is our product, and it runs both Antigravity and Claude Code on each lab’s plans. We checked every plan, model list, credit rule and price below against Google’s and Anthropic’s own pages, and the coding scores against Artificial Analysis’s, on November 20, 2026, and we say where you don’t need Enjoy at all, which is most of the time.

Google AI Ultra vs Claude Max, compared

Both labs sell a $100 plan and a $200 plan for people who outgrow $20 a month, and both have recently renamed or reshuffled what sits in each tier. Anthropic now calls its $100 tier Max and its $200 tier Ultra (“The Max 20x plan is now called the Ultra plan,” says its Ultra page). Google sells Google AI Ultra 5x at $99.99 and Google AI Ultra 20x at $199.99, and since November 18 its newest coding model, Gemini 4 Neon, is Antigravity’s default on $19.99 Pro as well as Ultra. That changes the math before you spend $100.

Plan Best for Price, checked November 20, 2026 Weekly usage Five-hour limit Coding agent and top models
Google AI Ultra 5x Running out of Pro in Antigravity, Gemini 4 Carbon, and Google’s bundle $99.99 a month 5 times Pro’s tokens in Antigravity, per Google’s May post Yes, refreshed every five hours Antigravity: Gemini 4 Neon, Carbon, Orchestrator (Carbon), Claude Opus 5.5
Claude Max Claude Code with Fable 5.2, at $100 $100 a month 5 times Claude Pro Yes, 5 times Claude Pro Claude Code: Opus 5.5; Fable 5.2 up to half the week
Google AI Ultra 20x All-day Antigravity use, Project Genie, and 30 TB of storage $199.99 a month 20 times Pro’s tokens in Antigravity, per Google’s May post Yes, refreshed every five hours As Ultra 5x
Claude Ultra Long Claude Code sessions with no five-hour stop $200 a month 10 times Claude Pro None once verified As Claude Max
Two plans, one from each lab Both top coding setups for less, or two separate limits $39.99 for both Pro plans; $119.99 for one $100 plan plus the other’s Pro; $199.99 for Max plus Ultra 5x Each plan’s own Each plan’s own Both Antigravity and Claude Code
Enjoy, with any of these Watching both meters and moving one conversation across agents A monthly subscription on top of your plans, free for 7 days (pricing) Your plans’ Your plans’ Antigravity (Gemini models), Claude Code and other labs’ agents

Prices are each company’s US list prices before tax, and all four $100 and $200 tiers bill monthly only. Google’s 5x and 20x labels come from its subscriptions page (“5x higher usage limits vs. AI Pro” and “20x higher usage limits vs. AI Pro”), where its Google One plan table places those multiples under the Gemini app and rates Antigravity agent requests only as “Expanded” on Pro, “Higher” on Ultra 5x and “Highest” on Ultra 20x. For Antigravity itself, Google’s May 19 post on its plans gives the same numbers: “The $100/month Ultra plan has a rate limit of 5X worth of tokens compared to the $20/month Pro plan, and the $200/month Ultra plan has a rate limit of 20X worth of tokens compared to the $20/month Pro plan.”

How the models and limits compare

On Artificial Analysis’s Coding Agent Index 2.1, checked November 20, 2026, the two labs’ best coding setups are one point apart. The surprise is which plan unlocks which model:

  • Antigravity CLI with Gemini 4 Neon (xhigh) scores 59 (43 on FrontierCode-AA 2.0, 65 on Terminal-Bench 4.1, 68 on SWE-Atlas-QnA-AA), averaging $6.86 and 30.5 minutes a task in API terms. Neon is included on paid $19.99 Google AI Pro, not gated to Ultra.
  • Antigravity CLI with Gemini 4 Carbon (high), the model gated to Google AI Ultra among personal plans, scores 55 (38 / 65 / 64), at $9.75 and 33.6 minutes a task. Google built Carbon for deep reasoning, cybersecurity and open-ended research rather than everyday speed.
  • Claude Code with Claude Fable 5.2 (high) scores 58 (49 / 64 / 62), at $7.51 and 31.3 minutes a task, and leads FrontierCode-AA 2.0 by six points over Neon. Fable 5.2 is included on Claude Max and Ultra; on Claude Pro it bills usage credits after November 16 (our memo).
  • Claude Code with Claude Opus 5.5 (max) also scores 58 (44 / 65 / 66), at $12.61 and 1.1 hours a task, and is already the default on $20 Claude Pro.

Those costs are pay-per-token API prices, not subscription prices; on a plan, a task draws from your quota instead. Beyond the benchmark scores, the four plans diverge sharply on limits, credits and what else comes in the box:

Google AI Ultra 5x and 20x Claude Max and Ultra
Five-hour limit Yes, on both 5x and 20x Yes on Max; none on verified Ultra
Multiple of Pro stated 5 times Pro’s tokens on 5x; 20 times on 20x (May post); Claude models on a separate fixed limit 5 times Pro on Max; 10 times Pro’s weekly usage on Ultra
Exclusive top model Gemini 4 Carbon and Orchestrator (Carbon) Claude Fable 5.2 included (up to 50% of weekly limit)
Other lab’s models inside Claude Sonnet 5.5 and Opus 5.5 in Antigravity (no Fable) Claude models only
Monthly API / Cloud credits $40 (5x) or $100 (20x) in Cloud credits; stack and last 12 months; $5 prepayment required $100 (Max) or $200 (Ultra) in Claude API credits; expire each cycle; no payment method needed
Credits usable for main coding agent? Not for Antigravity overages (usable via Gemini API in AI Studio, OpenCode or Antigravity CLI with an API key) Not for interactive Claude Code or extra plan usage (usable for API, Agent SDK and claude -p with a Console API key)
Overage past limits Purchased Google One AI credits (not in Japan or via third-party billing) Usage credits at API rates (credits bought from Dec 7, 2026 expire in 12 months)
Non-coding bundle 20 or 30 TB storage, YouTube Premium Family, 1,000 or 2,000 Colab units, Home & Health Premium, Project Genie on 20x Claude apps, Claude Cowork, priority access at high traffic times
Account eligibility Personal Google Accounts only Personal accounts (Team/Enterprise sold separately)

Our Antigravity vs Claude Code guide compares the two agents themselves.

Google AI Ultra 5x

Best for: developers already hitting Google AI Pro’s weekly limit in Antigravity, anyone who wants Gemini 4 Carbon, or anyone who would pay for 20 TB of Google storage and a YouTube Premium family plan anyway.

Ultra 5x gives you Antigravity’s full model list, adding Gemini 4 Carbon (“our strongest model”) and Orchestrator (Carbon) to Neon, Argon, Hydrogen and Claude Sonnet and Opus 5.5, with what Google’s subscriptions page calls “Higher rate limits to agent model in Google Antigravity, our agentic development platform.” Outside Antigravity, Google One’s plan table bundles 20 TB of storage, a YouTube Premium Family plan in over 100 countries, 1,000 monthly Google Colab compute units, Google Home Premium Advanced, Google Health Premium, and $40 a month in Google Cloud credits through the Google Developer Program. Those Cloud credits can be claimed in Google AI Studio and spent on the Gemini API, and each month’s grant lasts 12 months and stacks (our guide to claiming them).

Price: $99.99 a month. “$99.99/mo - Ultra plans are billed monthly only,” says Google One.

Watch for: Ultra 5x’s five times Pro is in tokens, from a May post that today’s plans page doesn’t repeat, and that page notes “Usage limits for this service are subject to modification.” Claude’s models in Antigravity stay on “a separate fixed rate limit,” which Google hasn’t said grows with the plan. Ultra 5x still enforces both a five-hour refresh and a weekly limit. It is also personal-only: “Only personal Google Accounts can sign up for Google AI plans,” and “Google Workspace plans don’t include Antigravity plans or higher rate limits.” To spend the $40 monthly Cloud credits on the Gemini API, “Gemini API only applies free credits when that account has some of your own money in it. $5 is the minimum,” and those credits don’t pay for Antigravity subscription overages. Overages use separate Google One AI credits, which aren’t sold in Japan or on Apple, Pixel Pass or third-party billing, per Google One’s help pages.

Claude Max

Best for: developers who want Claude Code as their main agent at $100, with Fable 5.2 included and a stated five-times jump over Pro.

“The Max plan includes five times the Pro plan’s usage every five hours and every week,” says Anthropic’s Max page, alongside “Fable included, up to half your weekly limit,” “Priority access at high traffic times,” and “$100 a month in API credits to build on the Claude Platform.” On Claude Code, Opus 5.5 is the default and /model fable switches to Fable 5.2 (requiring Claude Code v2.1.321 or later, per its model docs), and if you hit the five-hour limit mid-turn it may briefly finish the step (“Usage limit reached · wrapping up”), counting toward the weekly limit. Unlike Google’s Cloud credits, Anthropic’s API credits page says “No payment method needed” on the Claude Platform once you’ve been on the plan for seven days, and the credits cover the API, Agent SDK, Managed Agents and claude -p run with a linked Console API key.

Price: $100 a month. “The Max plan is currently available as a monthly subscription only.”

Watch for: three separate limits. “Max plans have a five-hour limit, a weekly usage limit that applies across all models, and a separate weekly limit for Fable,” and Anthropic’s Claude Code guide states “These limits are shared across Claude and Claude Code, meaning all activity in both tools counts against the same limits.” Fable’s 50% allowance isn’t extra headroom: “You can use up to 50% of your weekly usage limits on Fable models at no extra cost. They draw from your plan’s regular weekly usage limits and use them faster than other Claude models,” says Anthropic’s Fable page. The $100 API credits don’t roll over (“Unused credits expire at the end of each billing cycle”) and Anthropic’s platform docs exclude “Interactive Claude Code in the terminal, IDE, desktop, or web; Extra usage in Claude, Claude Code, or Claude Cowork.” Going past your plan limits in Claude Code requires buying separate usage credits at API rates ($4/$20 per million input/output tokens for Opus 5.5, $8/$40 for Fable 5.2).

Google AI Ultra 20x

Best for: people who run Antigravity agents hard enough to hit Ultra 5x’s limits most weeks, want $100 a month of rolling Cloud credits, or specifically want 30 TB of storage and Project Genie.

Ultra 20x has the exact same Antigravity models as Ultra 5x, stepping up from “Higher” to “Highest rate limits to agent model in Google Antigravity.” In Google One’s table, storage rises from 20 TB to 30 TB, Colab compute units double from 1,000 to 2,000 a month, monthly Cloud credits rise from $40 to $100 (still stacking for 12 months each), and footnote 10 notes Project Genie is “Only available with a Google AI Ultra 20x plan.”

Price: $199.99 a month, billed monthly only.

Watch for: unlike Claude Ultra at $200, Google AI Ultra 20x does not remove the five-hour limit. Antigravity’s plans page describes Ultra as “The highest, most generous quota, refreshed every five hours” and “The highest weekly rate limits,” and by Google’s May post, 20x has four times 5x’s tokens, for twice the price. At $199.99 a month, Ultra 20x costs the same as pairing Ultra 5x ($99.99) with Claude Max ($100), so step up to 20x only when Ultra 5x is actually stopping you or you know you’ll use the extra Cloud credits, storage or Project Genie.

Claude Ultra

Best for: developers who run long, uninterrupted Claude Code sessions where a five-hour cap breaks their day, or who routinely burn through a full Max week.

“Ultra includes 10 times the Pro plan’s usage every week and has no five-hour limit, for verified accounts,” says Anthropic’s Ultra page. Everything else matches Max, with “$200 a month in API credits, up from $100 on Max.” Removing the five-hour window is the cleanest advantage either $200 plan offers for coding: you can spend your weekly budget in two heavy days instead of pacing around a five-hour clock.

Price: $200 a month, billed monthly only.

Watch for: verification isn’t guaranteed on day one. “If your account isn’t verified yet, a temporary limit applies,” says Anthropic, while it checks factors such as account and payment history, without publishing a timeline. Fable 5.2 still has its separate 50%-of-week cap drawn from the shared weekly pool, and the $200 in API credits still expire at the end of every billing cycle and can’t pay for interactive Claude Code. If you don’t regularly hit Max’s five-hour wall or weekly cap, Ultra gives you the same models at twice the price.

Two plans, one from each lab

Best for: getting both Gemini 4 Neon and Claude Code on separate limits, at $40, $120 or $200 a month instead of betting everything on one vendor.

Because Google put Gemini 4 Neon on paid $19.99 Pro and Anthropic puts Opus 5.5 on $20 Pro, splitting your budget is stronger here than in almost any other coding-plan choice:

  1. Google AI Pro + Claude Pro ($39.99/month, or $36.99 with annual Claude Pro): gets you Antigravity with Neon and Claude Code with Opus 5.5 for less than half the price of a single $100 plan, with two independent five-hour and weekly meters.
  2. Claude Max + Google AI Pro ($119.99/month): adds Fable 5.2 and 5x Claude Code usage, while keeping Neon in Antigravity for $20 as a fast second model and backup meter.
  3. Google AI Ultra 5x + Claude Max ($199.99/month): matches the price of one $200 plan while giving you both Carbon and Fable 5.2, 20 TB + YouTube Premium Family, $40/month rolling Google Cloud credits, and $100/month Claude API credits.

One instruction file works across both agents. Claude Code’s memory docs note “Claude can also read a repository’s AGENTS.md files in place of CLAUDE.md,” and Antigravity’s rules docs say “You can place an AGENTS.md or GEMINI.md file—or a .agents/rules/ directory—in your workspace folder and in any parent folder up to the repository root.”

Price: $39.99, $119.99 or $199.99 a month depending on the pair.

Watch for: Google’s FAQ warns “Using third-party software, tools, or services to access Antigravity is a violation of our Terms of Service” and “Wrapping agy inside third-party coding agents or applications, or automating account rotation, is not supported under consumer plans.” Use each plan in its own agent (or through Google’s supported ACP server), and expect to carry context between conversations yourself unless an app bridges them. Also remember Google AI Pro free trials exclude Claude 5.5 and only include Neon on eligible devices, so start a paid Pro subscription rather than a trial.

Enjoy, with any of these

Best for: running Antigravity and Claude Code side by side, watching both limits in one place, and handing a live conversation from one agent to the other when a plan runs out.

Enjoy isn’t a Google or Anthropic subscription: it’s a desktop app for Mac, Windows and Linux that runs Antigravity and Claude Code, alongside Codex, Grok Build, Muse Code and local models through OpenCode, on the plans you already pay for. What it adds to this choice:

  • Both meters, side by side. Enjoy’s usage meters display each signed-in account’s remaining quota and reset times together, so you can see whether a $100 or $200 plan is actually necessary.
  • One conversation across both labs. You can switch agents mid-thread: “The new agent gets the conversation so far, including replies from other agents, before it works on your message.”
  • Fewer stops at limits. When an agent hits its limit, Enjoy can switch to another account on supported agents or wait for the reset and pick up where it stopped.
  • Downgrade checks on Claude. On Claude Max, Enjoy’s plan suggestions check your last four weeks and tell you when a smaller plan would have covered your work.

Price: Enjoy is one monthly subscription, and the first seven days are free, with a card to start. It includes your phone and any browser and one computer; each other computer is $9.99 a month, and each teammate you invite needs a subscription of their own. See pricing.

Watch for: Enjoy doesn’t expand either lab’s limits or put Claude on a Google account. Antigravity in Enjoy runs through Google’s ACP server (“for Xcode, JetBrains IDEs, Zed, and custom ACP clients,” per Antigravity’s changelog), not by wrapping agy, and runs “the Gemini models Google offers your account,” as our Antigravity guide notes, so if you want Claude inside Antigravity, use Google’s own app, or run Claude Code on a Claude plan. Antigravity also uses one Google account per machine, so Enjoy can’t rotate across multiple Google accounts the way it does for Claude Code, Codex and Grok Build. Enjoy runs Claude Code headless, so if the reports in issue #97074 (still open as of November 20, 2026) affect you, Claude Code in Enjoy can burn Max’s five-hour window faster than in a terminal; that doesn’t affect a verified Claude Ultra account, which has no five-hour window. Plan suggestions cover Claude Code only and still display the older Max 5x and Max 20x labels. Waiting for a reset and scheduled workflows only run while your computer is awake with Enjoy open. Enjoy isn’t open source, and it’s a separate monthly subscription once the seven-day trial ends.

When to stay on the plan you have

  • Stay on Google AI Pro ($19.99) or Claude Pro ($20) if you run out only a couple of times a month. Paid Google AI Pro already includes Gemini 4 Neon and Claude Opus 5.5, and Claude Pro already includes Opus 5.5. Both labs sell pay-as-you-go overages (Google One AI credits on Pro, Claude usage credits on Pro) that cost less than jumping to $100 for the occasional heavy afternoon. Our Google AI Pro vs Ultra and Claude Pro vs Max vs Ultra guides break down when an upgrade pays off.
  • Stay on your current $100 or $200 plan if its agent already handles your codebase reliably. One point on a benchmark index doesn’t beat knowing how your primary agent behaves on your repos, and because all four tiers bill monthly, you can test the other lab for a month whenever your current plan stops fitting.
  • Drop from $200 to $100 if you haven’t hit a weekly limit in the last month. On Google, Ultra 5x has the exact same Antigravity models as Ultra 20x; on Anthropic, Max has the same models as Ultra, and the only coding features you give up are the second half of the weekly cap and the removal of the five-hour window.

How we compared

We read Google’s subscriptions page, Google One’s AI plans comparison table, Antigravity’s pricing, plans, models, FAQ, rules and changelog pages and its May post on changes to its plans, Google One’s AI credits help pages, and Google AI Studio’s Cloud credits guide; Anthropic’s pricing page, its help center articles on Max, Ultra, Claude Code limits, Fable models, monthly API credits and usage credits, its Claude Platform API credits docs, Claude Code’s model and memory docs, and Claude Code GitHub issue #97074; and Artificial Analysis’s Coding Agent Index 2.1 comparison of Antigravity CLI and Claude Code, all on November 20, 2026, and every page is linked below. We did not run our own quota tests. Because neither Google nor Anthropic publishes the size of a plan’s quota, only multiples of its Pro plan, we quote each vendor’s own wording and leave unverified figures out. Plans, models and limits 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.

11.Questions.

Is Google AI Ultra or Claude Max better for coding?

Claude Max for Claude Code, Fable 5.2 and clearer usage rules; Google AI Ultra for Antigravity, Gemini 4 Carbon alongside Claude Opus 5.5, and non-coding perks. On Artificial Analysis’s Coding Agent Index 2.1, Antigravity with Gemini 4 Neon scores 59 and Claude Code with Fable 5.2 or Opus 5.5 scores 58, so the top setups are close, and Neon is already on $19.99 Google AI Pro.

What’s the difference between Google AI Ultra 20x and Claude Ultra at $200?

The biggest coding difference is the five-hour limit. Anthropic’s Claude Ultra, formerly Max 20x, has 10 times Pro’s weekly usage and “no five-hour limit, for verified accounts.” Google AI Ultra 20x still has a five-hour limit alongside its weekly limit, adds Gemini 4 Carbon and Project Genie, and bundles 30 TB of storage, YouTube Premium Family and $100 a month in Cloud credits that last 12 months.

Does Google AI Ultra include Claude models for coding?

Yes, in Antigravity, up to Opus 5.5, but not Fable 5.2 and not Claude Code. Google’s models page includes Claude Sonnet 5.5 (thinking) and Claude Opus 5.5 (thinking) on paid Google AI Pro and Ultra. If you want Claude Fable 5.2 or Anthropic’s own agent, you need a Claude plan.

Which plan gives more coding usage, Google AI Ultra or Claude Max?

Neither company’s pages can tell you, because each states multiples of its own Pro plan, whose size neither publishes. Anthropic says Max has five times Pro’s five-hour and weekly usage and Ultra has 10 times Pro’s weekly usage. Google’s Antigravity blog said in May that Ultra 5x has “5X worth of tokens” and Ultra 20x “20X worth of tokens” compared to Pro, with Claude’s models in Antigravity on “a separate fixed rate limit.”

Do the monthly API or Cloud credits pay for Antigravity or Claude Code?

Not for interactive sessions on your plan. Anthropic’s $100 or $200 in monthly API credits expire each billing cycle and don’t cover interactive Claude Code or extra plan usage. Google’s $40 or $100 in monthly Cloud credits last 12 months and stack, after a $5 prepayment, and pay for the Gemini API in AI Studio, OpenCode or Antigravity CLI with an API key, not Antigravity subscription overages.

Should I buy two smaller plans instead of Google AI Ultra or Claude Max?

Often, yes. Paid Google AI Pro and Claude Pro cost $39.99 a month together, under half of either $100 plan, and already include Gemini 4 Neon in Antigravity and Opus 5.5 in Claude Code, with two separate sets of limits. For $199.99, the price of one $200 plan, you can have both Google AI Ultra 5x and Claude Max, and Enjoy hands one conversation between the two agents.

Enjoy Agentic Industries · Checked against its sources on