ChatGPT Pro vs Claude Ultra vs Google AI Ultra: Complete Comparison and Report on the $200 Tier, Usage Limits, Flagship Models, and API Credits
ChatGPT Pro, Claude Ultra and Google AI Ultra are the top consumer subscriptions of OpenAI, Anthropic and Google, and in November 2026 all three have a tier priced at about $200 a month. OpenAI sells Pro at $100, $200 or $500 a month, with Pro reasoning powered by GPT-6 Astra. Anthropic created Ultra in late October 2026 by renaming the former Max 20x plan, kept the $200 price, and gives it 10 times the Pro plan's weekly usage. Google split AI Ultra into two tiers at Google I/O on May 19, 2026: a 5x tier at $99.99 and a 20x tier at $199.99, after cutting the top tier from $250.
The verdict depends on what the $200 has to buy. ChatGPT Pro at $200 and Claude Ultra both return $200 a month in API credits, so a subscriber who spends that budget on the API pays nothing net for the app itself, while Google AI Ultra 20x includes $100 in Google Cloud credits and wins on everything outside the chat window: 30 TB of storage, YouTube Premium, 25,000 Flow credits, Deep Think and Project Genie. On independent measurements of the flagship models included in each plan, Claude Opus 5.5 scores 59 on the Artificial Analysis Intelligence Index v5.0.1 at max effort, GPT-6 Astra scores 57 at max effort and Gemini 4 Argon scores 50 at high effort.
For heavy coding and agent work inside the app, Claude Ultra is the plan with the clearest published rules: no five-hour limit for verified accounts, one weekly limit shared between Claude and Claude Code, and Fable models allowed up to 50% of that limit. ChatGPT Pro is the choice for people who want OpenAI's top reasoning model, no five-hour limit in Codex and an always-on agent (Dot), while Google AI Ultra 20x is the better buy for a household or creator that values storage, video generation and Google app integration as much as the chatbot. Prices in this comparison were verified on the official plan pages on November 8, 2026.
··········
CHATGPT PRO, CLAUDE ULTRA AND GOOGLE AI ULTRA: RELEASE TIMELINE AND PLAN IDENTITY.
The three plans reached their current shape at different moments in 2026, and each vendor sells its $200 tier as part of a ladder rather than as a single product. Google moved first: its May 19, 2026 subscription post introduced "a $100 AI Ultra plan" for developers, technical leads, knowledge workers and creators, and reduced "the monthly price of our top-tier AI Ultra plan from $250 to $200." The same post switched Gemini app limits from daily prompt caps to a compute-based model that refreshes every five hours until a weekly limit is reached. Google's current subscription page lists the two tiers at $99.99 and $199.99 a month, and Google One help pages name them Google AI Ultra 5x and Google AI Ultra 20x.
Anthropic's Ultra is the youngest of the three names. In late October 2026 the Max 20x plan became Ultra at the same $200 price, and the Ultra help article states that verified Ultra accounts no longer have a five-hour limit. Max remains on sale at $100 a month with 5 times Pro usage, so Anthropic's ladder mirrors Google's: one tier near $100 and one near $200.
OpenAI's Pro plan carries three usage tiers inside one plan name. The ChatGPT pricing documentation lists "Plans at $100, $200, or $500 USD per month," and the API credits documentation calls them the Pro $100, Pro $200 and Pro $500 tiers. The model behind Pro reasoning, GPT-6 Astra, was released in September 2026 according to Artificial Analysis.
........
Item | ChatGPT Pro | Claude Ultra | Google AI Ultra |
Vendor | OpenAI | Anthropic | |
Tier compared | Pro $200 | Ultra | Ultra 20x |
Monthly price (US) | $200 | $200 | $199.99 |
Sister tier near $100 | Pro $100 | Max ($100, 5x Pro) | Ultra 5x ($99.99) |
Higher tier | Pro $500 (Ultrafast) | None for individuals | None |
Current structure since | Three Pro tiers on the pricing page | Late October 2026 (Max 20x renamed Ultra) | May 19, 2026 (5x and 20x tiers) |
Annual billing | Not listed | Monthly only | Monthly only |
Flagship model in the app | GPT-6 Astra (Pro reasoning) | Claude Opus 5.5, Fable 5 and 5.1 | Gemini 4 Argon, Deep Think |
........
··········
THE $200 PRICE IN DETAIL: WHAT EACH PLAN CHARGES AND WHAT IT RETURNS.
At list price the three plans cost the same within one cent, and none offers an annual discount at this tier. Anthropic's help article says Ultra is "currently a monthly subscription only," and the Google One calculator states that "Ultra plans are billed monthly only." OpenAI's pricing pages list no annual price for Pro. Anthropic also notes that Ultra's $200 applies to web subscriptions and that mobile pricing may differ, a caveat relevant to buyers who subscribe through the App Store or Google Play.
The difference appears in what comes back to the subscriber each month. OpenAI's API credits guide lists $200 of monthly API credits for the Pro $200 tier, usable across the Responses, Agents, Realtime and Images APIs and built-in tools, but not for Codex usage through API keys. Anthropic's documentation lists $200 a month for Ultra, usable on the Claude API, Claude Managed Agents, the Claude Agent SDK and the Console Playground. Google's AI plans page lists $100 a month in Google Cloud credits for the 20x tier (and $40 for the 5x tier), plus 2,000 Google Colab compute units.
........
Monthly value | ChatGPT Pro $200 | Claude Ultra | Google AI Ultra 20x |
Subscription price | $200 | $200 | $199.99 |
Developer credits included | $200 OpenAI API credits | $200 Claude API credits | $100 Google Cloud credits |
Other compute | None listed | None listed | 2,000 Colab compute units |
Credits roll over | No, expire at end of billing cycle | No, expire at end of billing cycle | Not stated on plan page |
Waiting period | Subscription at least 7 days old | Claim after 7 days on the plan | Not stated |
Extra app usage | Purchasable credits | Usage credits at standard API rates | Google AI credits (Flow, Antigravity) |
Annual cost | $2,400 | $2,400 | $2,399.88 |
........
Both OpenAI and Anthropic keep the API credits separate from app usage. Anthropic's Claude Code help page states that Ultra's monthly API credits "can't be used for Claude Code, Claude apps, or usage credits," and OpenAI says Pro credits do not cover Codex used through API keys. The credits are a developer benefit, worth their full value only to someone who already builds on the API.
··········
USAGE LIMITS ON CHATGPT PRO, CLAUDE ULTRA AND GOOGLE AI ULTRA.
Each vendor describes its top-tier limits in its own unit, and only Anthropic and Google publish a multiplier against a named lower plan. Claude Ultra provides "10 times the Pro plan's usage every week," with a weekly limit across all models and no five-hour session limit once the account is verified. Verification is automatic and based on account and payment history; Anthropic says verifying a phone number at checkout does not remove the temporary limit, and a missing "Current session" bar under Settings > Usage shows that the account is verified.
Google AI Ultra 20x offers "20x higher usage limits vs. AI Pro," and the Gemini Apps help page describes the Ultra tiers as "5x or 20x higher than AI Pro limits depending on your subscription." Those limits are compute-based, depend on prompt complexity, features and chat length, and refresh every five hours until the weekly limit is reached. Google AI Pro is in turn rated at 4x the standard free limits, so the 20x tier sits at 80 times the free baseline by Google's own multipliers.
OpenAI describes ChatGPT Pro in qualitative terms. The pricing page promises "Unlimited messages and uploads," subject to abuse guardrails, and lists GPT-5.6 Sol, Terra, Luna and GPT-5 Thinking Mini as unlimited, while GPT-6 Astra, GPT-6 Sol, GPT-6.1 Sol and GPT-6 Luna are marked "Expanded." The Codex documentation states that "Pro plans currently have no five-hour limit" and that weekly limits may apply. The per-tier comparison of the $100, $200 and $500 Pro plans sits on an OpenAI help article that returned an error on November 8, so the exact multiple of the $200 tier over the $100 tier could not be verified.
........
Limit | ChatGPT Pro $200 | Claude Ultra | Google AI Ultra 20x |
Published multiplier | Not published on reachable pages | 10x Claude Pro, weekly | 20x Google AI Pro |
Five-hour window | None in Codex ("no five-hour limit") | None for verified accounts | Refreshes every five hours |
Weekly limit | "Weekly limits may also apply" | Yes, across all models | Yes, compute-based |
Model-specific cap | Expanded access to GPT-6 models | Fable up to 50% of weekly limit | Highest access to Gemini 4 Argon |
Chat and coding tool share one limit | Usage shared with agentic features | Yes, Claude and Claude Code | Gemini app and Antigravity both 20x |
Overage option | Buy credits | Usage credits, auto-reload on web | Google AI credits, up to 5 auto-reloads a day |
........
A practical reading of the table: Claude Ultra and Google AI Ultra state their ceilings as multiples a subscriber can test against a cheaper plan, and Claude Ultra removes the five-hour wait outright for verified accounts. ChatGPT Pro offers the broadest "unlimited" wording but ties it to abuse guardrails and to models from the GPT-5.6 generation, while access to the newest GPT-6 models is described only as expanded.
··········
FLAGSHIP MODELS AND CONTEXT WINDOWS IN EACH $200 PLAN.
The models a subscriber reaches through the app decide most of the day-to-day difference. ChatGPT Pro lists "Pro reasoning powered by GPT-6 Astra" and is the only plan that includes GPT-5.6 Sol Pro. Claude Ultra includes Sonnet, Haiku and Opus, plus Fable 5 and Fable 5.1 as "a standard part of your plan," with up to 50% of weekly limits usable on Fable at no extra cost. Google AI Ultra gives the "highest access to Gemini 4 Argon," available in the Gemini app to Pro and Ultra subscribers since October 13, 2026, and Deep Think, which Google's help page reserves for Google AI Ultra subscribers and describes as an experimental capability running on Gemini Pro.
Context windows are published differently. OpenAI gives Pro a 128K total context in Instant mode (about 250 pages of input) and 400K in reasoning mode (about 680 pages), against 54K and 256K on Plus. Anthropic lists a context window of up to 1M tokens on all plans, varying by model. Google's Gemini Apps help page lists 1 million tokens for both AI Pro and AI Ultra, and the Google One Ultra benefits page adds file uploads of up to 1,500 pages and up to 600 sources per Gemini Notebook.
........
Model and context | ChatGPT Pro $200 | Claude Ultra | Google AI Ultra 20x |
Top reasoning model | GPT-6 Astra | Claude Opus 5.5 | Gemini 4 Argon |
Other premium models | GPT-6.1 Sol, GPT-6 Sol, GPT-5.6 Sol Pro | Fable 5, Fable 5.1 | Deep Think (Ultra only), Nano Banana |
Context in the app | 128K Instant, 400K reasoning | Up to 1M, varies by model | 1M tokens |
Input estimate | About 250 pages (Instant), 680 pages (reasoning) | Not stated in pages | Up to 1,500-page file uploads |
Fastest mode | Ultrafast only on Pro $500 | Priority access at high traffic | Not stated |
........
One gap readers often ask about concerns the top of the leaderboards. Gemini 4 Carbon leads the Artificial Analysis Intelligence Index at 60, yet it is limited to a pre-release program and is not part of Google AI Ultra, and OpenAI's GPT-6 Cyber is a trusted-access model outside ChatGPT Pro. Paying $200 buys the best generally available model of each vendor, not the restricted ones.
··········
INDEPENDENT EVALUATIONS OF GPT-6 ASTRA, CLAUDE OPUS 5.5 AND GEMINI 4 ARGON.
No independent organization measures consumer plan allowances, so the closest neutral comparison is the score of each plan's flagship model. Artificial Analysis runs every model through the same Intelligence Index v5.0.1, which combines AA-Briefcase v1.1, AutomationBench-AA, Terminal-Bench 4.0, FrontierCode 2.0, AA-Omniscience v1.1, GDP.pdf, HLE-Diamond, CritPt v2 (Hard Subset) and Terminal-Bench-Science 0.1. Its comparison pages, read on November 8, 2026, give the results below; each figure carries the reasoning setting Artificial Analysis used.
........
Model (Artificial Analysis setting) | Intelligence Index v5.0.1 | Output speed | Context | Included in |
Claude Opus 5.5 (Max, default fallback) | 59 | 114 tok/s | 1M | Claude Ultra |
Claude Opus 5.5 (Xhigh, default fallback) | 57 | Not on page | 1M | Claude Ultra |
GPT-6 Astra (Max) | 57 | 42 tok/s | 1M | ChatGPT Pro |
Gemini 4 Argon (High) | 50 | 71 tok/s | 1M | Google AI Ultra |
Claude Fable 5.1 (Max, default fallback) | 49 | 87 tok/s | 1M | Claude Ultra |
........
Three cautions apply. The context and speed figures describe the models as Artificial Analysis measured them through the API, which is not the same as the app: ChatGPT Pro, for example, documents a 400K reasoning context in the app, below the 1M API figure. The index measures the model at a chosen effort level, and the apps do not always run at max effort by default; Anthropic's Opus 5.5 announcement gives medium as the default effort. And Artificial Analysis lists Gemini 4 Argon at "High," the setting it tested, while Deep Think in the Gemini app is a separate mode that the index does not score.
··········
VENDOR-PUBLISHED BENCHMARKS THAT ONLY ONE COMPANY REPORTS.
Each vendor publishes its own figures on benchmarks the others do not report, which makes them useful for understanding a model's strengths but not for ranking the plans. Anthropic's Opus 5.5 announcement of September 22, 2026 reports 66.4% on Terminal-Bench 4.0 at xhigh effort, 81.8% on OSWorld 2.1 (partial credit, max effort), 67.7% on Humanity's Last Exam with tools at max effort, and an Elo of 1846 on GDPval-AA v2.1 at max effort. Google's October 13, 2026 Gemini 4 Argon post reports 87% on ERQA embodied reasoning and 57% on RareBench and 52.2% on LabBench, the last two from private Gamow Labs evaluations, and says Argon ranks first on the Vals Index without giving a score.
OpenAI's GPT-6 Astra announcement page and its help article on GPT-6 models in ChatGPT both returned errors on November 8, so this comparison uses no OpenAI-published Astra benchmark. The Artificial Analysis score above is the only Astra performance figure used here.
··········
[[ADPLUS_300x250_1]]
CODING AND AGENT TOOLS: CODEX, CLAUDE CODE AND ANTIGRAVITY.
All three $200 plans bundle a coding agent, and the limits around those agents differ more than the chat limits do. Claude Ultra includes Claude Code, and Anthropic states that limits "are shared across Claude and Claude Code, meaning all activity in both tools counts against the same limits," IDE usage included. A verified Ultra account has only the weekly limit, and once it is reached the subscriber can enable usage credits billed at standard API rates.
ChatGPT Pro includes Codex and ChatGPT Work with "Longer Codex and ChatGPT Work sessions" than Plus, and the Codex documentation says Pro plans currently have no five-hour limit. Pro also adds Dot, which OpenAI calls "your always-on agent," with texting to a dot rolling out in beta to Pro users in the United States in late October 2026. OpenAI's changelog adds that GPT-6.1 Terra became available in ChatGPT Work and Codex on Plus, Pro and business plans in the week of October 26.
Google AI Ultra 20x gives the highest Antigravity rate limits, with a quota "refreshed every five hours" and "the highest weekly rate limits," plus access to third-party models inside Antigravity. When the quota runs out, Google AI credits are consumed at standard Gemini Enterprise consumption pricing. Google's Ultra page also lists Gemini Agent (US only, English only), first access to Tasks in Gemini in select countries and the highest Chrome auto browse access.
........
Agent feature | ChatGPT Pro $200 | Claude Ultra | Google AI Ultra 20x |
Coding agent | Codex, ChatGPT Work | Claude Code (terminal and IDE) | Google Antigravity |
Session limit | No five-hour limit | No five-hour limit (verified) | Five-hour refresh, weekly cap |
Always-on or background agent | Dot (texting beta, US) | Background tasks via Cowork in chat | Gemini Agent (US, English) |
Browser agent | Computer Use, Chrome (limited by region) | Claude in Chrome | Chrome auto browse (US, India) |
Overage billing | Purchased credits | Usage credits at API rates | AI credits at Gemini Enterprise rates |
........
··········
STORAGE, MEDIA AND BUNDLED SERVICES: WHERE GOOGLE AI ULTRA PULLS AHEAD.
Outside the chatbot, the Google plan carries far more than its two rivals. Google AI Ultra 20x includes 30 TB of cloud storage across Gmail, Drive and Photos, 25,000 Flow credits a month for video work with Veo 3.1, 30,000 monthly Flow Music credits, a YouTube Premium individual plan, Google Home Premium Advanced and Google Health Premium. Project Genie, the Genie 3 world model, is available only with Ultra, and Google's May post reserved it for the $200 tier.
ChatGPT Pro's media benefits stay inside ChatGPT: "Unlimited, faster image creation," unlimited voice and expanded projects, tasks and custom GPTs. Claude Ultra lists no storage, video or music benefit; its extras are priority access to new models, features and products and priority access at high-traffic times, which it shares with Max.
........
Bundled benefit | ChatGPT Pro $200 | Claude Ultra | Google AI Ultra 20x |
Cloud storage | Not included | Not included | 30 TB |
Image generation | Unlimited, faster | Not listed as a plan benefit | Nano Banana, highest access |
Video generation | Not stated on pricing page | Not included | Veo 3.1, 25,000 Flow credits |
Music generation | Not listed | Not included | 30,000 Flow Music credits a month |
Video streaming | Not included | Not included | YouTube Premium individual |
Smart home and health | Not included | Not included | Home Premium Advanced, Health Premium |
Early access | Research preview of new features | Priority access to new models and products | Project Genie (Ultra only) |
........
··········
WHAT THE INCLUDED CREDITS BUY: COST EXAMPLES FOR TWO WORKLOADS.
The API credits change the effective price of ChatGPT Pro and Claude Ultra only for subscribers who use them, so the arithmetic below states its assumptions. Prices are the official API list prices verified on November 8, 2026: GPT-6 Astra at $10 input and $50 output per 1M tokens, GPT-6.1 Sol at $2 and $10, Claude Opus 5.5 at $4 and $20, and Gemini 4 Argon at an introductory $2 and $10 that Google says rises to $4 and $20 after November 30, 2026. The calculations were run in Python.
The first workload is a developer's monthly agent pipeline of 20M input tokens and 2M output tokens, with no caching or batch discounts. On GPT-6 Astra it costs $300, so the $200 of Pro credits covers two thirds of it; on GPT-6.1 Sol it costs $60 and the credits cover it entirely. On Claude Opus 5.5 the same workload costs $120, inside the $200 of Ultra credits.
........
Workload: 20M input + 2M output tokens a month | List cost | Covered by plan credits | Out of pocket |
GPT-6 Astra (ChatGPT Pro $200 credits) | $300 | $200 | $100 |
GPT-6.1 Sol (ChatGPT Pro $200 credits) | $60 | $60 | $0 |
Claude Opus 5.5 (Claude Ultra credits) | $120 | $120 | $0 |
Gemini 4 Argon, intro price | $60 | Not stated for Cloud credits | Not computed |
Gemini 4 Argon, after November 30 | $120 | Not stated for Cloud credits | Not computed |
........
The second view asks how much a full month of credits buys at a 10-to-1 ratio of input to output tokens. The $200 OpenAI credit buys about 13.3M input and 1.3M output tokens on GPT-6 Astra, or about 66.7M input and 6.7M output tokens on GPT-6.1 Sol. The $200 Anthropic credit buys about 33.3M input and 3.3M output tokens on Opus 5.5. Google's plan pages describe the $100 as Google Cloud credits without saying which services they apply to, so no Gemini API figure is given for them.
........
Plan | Monthly price | Credit value | Net monthly cost if credits fully used | Net annual cost |
ChatGPT Pro $200 | $200 | $200 API | $0 | $0 |
Claude Ultra | $200 | $200 API | $0 | $0 |
Google AI Ultra 20x | $199.99 | $100 Cloud | $99.99 | $1,199.88 |
........
The net figures are a best case. Credits on both OpenAI and Anthropic expire at the end of each billing cycle, OpenAI requires the subscription owner to own the receiving API organization with the same email address, Anthropic lets credits be claimed only after 7 days on the plan, and they cannot pay for app usage. A subscriber who never opens the API console pays the full $200 for either plan.
··········
DATA, REGIONS AND ACCOUNT RULES ACROSS THE THREE PLANS.
Regional coverage is uneven, and several headline features are limited to the United States. Google says its AI plans are available in more than 140 countries and territories (the Ultra section of the same page says more than 150), but Gemini Agent is US and English only, Chrome auto browse is limited to the US and India, and many Ultra features require users to be 18 or older, including Antigravity, Project Genie and Deep Think. Google One members who subscribed through a third-party partner or Pixel Pass cannot upgrade to Ultra directly.
OpenAI marks Pro's composer predictions, memories, Computer Use, Chrome and Record & Replay as "Limited" with regional restrictions, and texting to a dot is a US-only beta. Anthropic's Ultra help article does not list regional limits, and its pricing page says prices exclude applicable tax and may change at Anthropic's discretion. All three vendors reserve the right to change limits: Google says Gemini limits "may change without notice, including due to capacity constraints," Anthropic says uninterrupted service is not guaranteed, and OpenAI ties unlimited use to abuse guardrails.
··········
WHAT OPENAI, ANTHROPIC AND GOOGLE DO NOT PUBLISH ABOUT THEIR $200 TIERS.
The evidence gaps fall unevenly. OpenAI's reachable pages do not state how much more usage the $200 Pro tier gives over the $100 tier, or the exact weekly limit for GPT-6 Astra; that comparison is on a help article that returned errors on November 8. Anthropic publishes the 10x multiplier and the 50% Fable rule but not an absolute message or token count for the weekly limit, and its Ultra help page carries no precise update date. Google publishes 5x and 20x multipliers and credit amounts but no prompt counts, no Deep Think quota and no statement of which services the $100 Google Cloud credits can pay for.
None of the three vendors publishes a per-plan figure that an independent organization has verified, which is why the model scores above come from the API versions of the models rather than from the subscriptions. A buyer testing a plan learns its practical ceiling only by using it, and each vendor shows remaining usage in its settings.
··········
WHICH $200 PLAN TO CHOOSE: CHATGPT PRO, CLAUDE ULTRA OR GOOGLE AI ULTRA.
The three plans cost the same and serve different buyers, so the choice follows the workload. Readers comparing the cheaper tiers can start from our guide to Claude plans and usage limits, the full list of ChatGPT plans and prices and our overview of Gemini plans from AI Plus to Ultra; the step below Ultra is covered in Claude Max vs Claude Ultra.
........
Workload or buyer | Recommended plan | Reason from the published facts |
Daily coding agent work, long sessions | Claude Ultra | No five-hour limit when verified, one weekly limit across Claude and Claude Code, Opus 5.5 scores 59 |
Developer who also spends on the API | ChatGPT Pro $200 or Claude Ultra | $200 of API credits each, net app cost $0 if fully used |
OpenAI's strongest reasoning model and an always-on agent | ChatGPT Pro $200 | GPT-6 Astra Pro reasoning, Dot, no five-hour limit in Codex |
Creator working with video, images and music | Google AI Ultra 20x | Veo 3.1 with 25,000 Flow credits, 30,000 Flow Music credits |
Household already on Google services | Google AI Ultra 20x | 30 TB storage, YouTube Premium, Home and Health Premium |
Very long documents in the chat app | Claude Ultra or Google AI Ultra 20x | Up to 1M context, against 400K for Pro reasoning |
Buyer who needs Ultrafast responses | ChatGPT Pro $500 | Ultrafast is limited to the $500 tier |
........
For a single subscriber choosing one $200 plan for professional work in November 2026, Claude Ultra is the strongest combination of published limits and model score, and ChatGPT Pro matches it on credit value while leading on OpenAI-specific features. Google AI Ultra 20x is the plan to pick when storage, media generation and Google's own apps carry as much weight as the assistant, accepting half the developer credit and a flagship model that scores 50 on the same index where Opus 5.5 scores 59.
··········
FOLLOW US FOR MORE.
·····
DATA STUDIOS
·····
datastudios.org
