Tools by Editorial Team

Paid AI Plans Compared: Claude Pro, Gemini, Copilot Pro

Claude Pro, Google AI Pro, and Microsoft 365 Copilot compared on price, model access, context, and limits, so you can pick the right $20/mo plan.

Three vendors sit at the $20/mo price point for a paid consumer AI assistant, and the differences between them are not just logos and model names. They differ on which underlying model you get, how much you can send it before the rate limit kicks in, whether the vendor trains on your conversations, and what other apps (Docs, Sheets, Outlook, the file on your desktop) get AI features along with the chat. This guide walks through what each plan costs, what each one unlocks over the free tier, and where each is the obvious pick.

TL;DR

  • The three contenders at the $20/mo floor are Claude Pro ($17/mo annual, $20/mo monthly), Google AI Pro (the Gemini Advanced bundle, listed in EUR on Google’s subscriptions page as EUR 21.99/mo), and Microsoft 365 Copilot (listed as three organisation SKUs: Microsoft 365 Copilot Business add-on at $18/user/mo annual or $25.20 monthly; Microsoft 365 Business Standard with Copilot at $23.50/user/mo annual or $28.20 monthly; Microsoft 365 Business Premium with Copilot at $32/user/mo annual or $38.40 monthly).
  • Model access differs in kind, not just in name. Claude Pro opens Anthropic’s full consumer lineup - Sonnet, Haiku, Opus, and Fable via usage credits - while Google AI Pro unlocks the Gemini 3 Pro path inside Search and the Gemini app, and Microsoft 365 Copilot layers GPT-5.X-priority access on top of Office.
  • Context windows, usage caps, and rate limits are stated differently. Claude advertises “Up to 1M” context on both Free and Pro with “at least 5x more usage per 5-hour session than Free” on Pro; Microsoft 365 Copilot emphasises “Work IQ” grounding across Microsoft 365 data plus priority access to “Latest OpenAI models: GPT-5.X” without a published per-user token cap.
  • Privacy is the obvious fourth axis, not a fifth. Claude Pro advertises no training on Pro or Max chat inputs by default (Anthropic’s consumer policy, separate from API). Google AI Pro inherits Google’s policy of not training on Gemini API or paid Gemini app inputs once Workspace data handling is on. Microsoft 365 Copilot inherits Microsoft 365’s enterprise data protection (“Copilot Chat is a web-grounded AI assistant available, at no additional cost, with eligible Microsoft 365 plans”). Each vendor’s own privacy page is the load-bearing reference.
  • Pick by what you already pay for, not by which model is “best”. If your work lives in Word, Excel, Outlook, and Teams, the right answer is Microsoft 365 Copilot. If you live in Google Docs, Gmail, and Search, the right answer is Google AI Pro. If you want the cleanest chat with the longest context and the strongest “no training on inputs” stance, the right answer is Claude Pro.

What you actually get for the $20/mo

Each vendor describes its paid tier in different language. Here is what the actual pricing pages say.

Claude Pro

The Claude pricing page lists five tiers: Free, Pro, Max, Team, and Enterprise. The Pro tier is positioned as “Pro plan. Great value for everyday use in larger codebases,” and the page quotes two prices side by side: “$17 Per month with annual subscription discount ($200 billed up front). $20 if billed monthly.” On the Free plan, only Sonnet and Haiku are accessible; Pro opens up “Opus, Fable, Sonnet Pro” via usage credits, adds Claude Code, Claude Science, Claude in Chrome and Microsoft 365, Research, and Projects, and raises the session cap to “at least 5x more usage per 5-hour session than Free.” Both Free and Pro advertise the same “Up to 1M” context window. The Max tier ($100/mo) is the 5x-or-20x option for users who hit Pro’s ceiling; Team models and Enterprise seat pricing are separate and out of scope for this consumer comparison.

Google AI Pro (Gemini Advanced)

The Gemini Subscriptions page is the consumer pricing surface, and it lists four tiers in EUR for Germany: a free tier with “Gemini 3.6 Flash” and “limited access to 3.1 Pro,” a EUR 4.99/mo “Google AI Plus” tier with “2x higher usage limits than free,” a EUR 21.99/mo “Google AI Pro” tier with “4x higher usage limits than free” and access to “Gemini 3 Pro model in Google Search,” and a EUR 99.99/mo “Google AI Ultra” tier with “Up to 20x higher usage limits” and “vor allen anderen Zugriff auf erweiterte Funktionen wie Deep Think und Gemini Spark.” The page explicitly notes “monatliche Abogebühr von derzeit 21,99 €” for AI Pro. The AI Pro tier is the consumer-equivalent of what most vendors call “Gemini Advanced” and what unlocks the higher-capability path through Gemini for AI Plus. AI Pro also unlocks “Google Home Premium (standard), YouTube Premium Lite, and 5 TB storage” alongside the AI access.

US dollar pricing for Google AI Pro is not published verbatim on the subscriptions page. USD figures for Google AI Pro and Google AI Ultra are typically reported in Google’s regional pricing summaries; this guide does not invent a USD number for Pro.

Microsoft 365 Copilot

The Microsoft 365 Copilot pricing page lists three organisation SKUs and one free chat tier. Microsoft 365 Business Premium with Copilot is $32.00/user/mo on the annual subscription (auto-renews) and $38.40/user/mo on the monthly subscription. Microsoft 365 Business Standard with Copilot is $23.50/user/mo annual and $28.20/user/mo monthly. The Microsoft 365 Copilot Business add-on is “Originally $21.00, now starting from $18.00 user/month, paid yearly” and $25.20/user/mo monthly, with the footnote “This discount offer is available between July 1, 2026, and December 31, 2026.” “Microsoft 365 Copilot Chat” is “$0” with metered usage for AI actions. The page documents that “A separate license for a qualifying Microsoft 365 plan is required to purchase Microsoft 365 Copilot. For up to 300 users.” For individuals, the consumer SKU lives on a separate product page that was not reachable for direct verification in this run; the Microsoft pricing page covers organisations only.

For organisations, the comparison table documents “Latest OpenAI models: GPT-5.X” with “Priority” access, “Model Choice” as a linked feature, “Work IQ” for grounding, and “Pre-built Microsoft agents, including Researcher, Analyst, and Facilitator.” The trial footnote notes “After your one-month free trial ends, your subscription will automatically convert to a paid subscription.” A footnote states “Microsoft Teams license required.” The discount footnote states “This discount offer is available between July 1, 2026, and December 31, 2026” for the Business add-on path.

The model inside each plan

The model is what actually answers your prompt. Here is what each vendor names and what it costs on the API side, so the plan-level choice has a comparable baseline.

Claude: Sonnet 5.5, Opus 5.5, Haiku 5.5, Fable 5.1

The Claude pricing page lists API pricing for four models alongside the consumer notion. The published rates are: Fable 5.1 at $10 input / $50 output per million tokens (with $0.25 cache read, $12.50 cache write); Opus 5.5 at $4 / $20; Sonnet 5.5 at $2 / $10; and Haiku 5.5 (under 100K context) at $0.10 / $0.50. The Claude Sonnet 5.5 product page describes Sonnet 5.5 as “featuring a 1M context window” with pricing at “$2 per million input tokens” and “$10 per million output tokens,” plus “90% cost savings with prompt caching” and “50% cost savings with batch processing.” The page also notes that “Pricing for Sonnet 5.5 will cost up to an estimated 30% less to run than Sonnet 5 for typical workloads billed by token.”

In the consumer product, the Claude pricing page states that Free users can use Sonnet and Haiku but not Opus or Fable; Pro users get Opus, Fable, and Sonnet via usage credits; Max users get “50% of weekly limits” of those models on the same terms, with the option to buy more. “Max plans at 50% of your weekly usage limit. Pro users can access Fable via usage credits” is the language used on the page. Free and Pro both list “Up to 1M” context, but Pro gets more usage per 5-hour window.

Gemini: 3.1 Pro Preview, 2.5 Pro, and the Flash tier

The Gemini API pricing page lists Gemini 3.1 Pro Preview at “$2.00, prompts <= 200k tokens / $4.00, prompts > 200k tokens” input and “$12.00, prompts <= 200k tokens / $18.00, prompts > 200k” output. Gemini 2.5 Pro is listed at “$1.25, prompts <= 200k tokens / $2.50, prompts > 200k tokens” input and “$10.00, prompts <= 200k tokens / $15.00, prompts > 200k” output. The Gemini API models page describes “Gemini 3 Flash (Preview)” as “Frontier-class performance rivaling larger models at a fraction of the cost,” and lists Gemini 3.5 Flash as “Our legacy Flash model, providing baseline speed and foundational performance for routine, high-throughput workloads.” The free tier provides “5,000 free search requests per month (shared across all Gemini 3.x models)” per the pricing page footnote, and “1,500 RPD (free)” on Gemini 2.5 Pro grounding.

In the consumer product, the Gemini Subscriptions page documents “Zugriff auf 3.6 Flash” and “Unterschiedlicher Zugriff auf 3.1 Pro” on the free tier, with AI Pro unlocking the Gemini 3 Pro path through Search. The page also documents “bis zu 1.500 Seiten lange Dateien hochladen” (upload files up to 1,500 pages) and “Code-Repository mit bis zu 30.000 Zeilen” (code repositories up to 30,000 lines) on AI Pro and AI Ultra tiers. AI Ultra adds “vor allen anderen Zugriff auf erweiterte Funktionen wie Deep Think und Gemini Spark,” which is what the page describes as the early-access tier for the higher-capability reasoning features.

Microsoft 365 Copilot: GPT-5.X with priority access

The Microsoft 365 Copilot pricing page names the underlying model family in the comparison table as “Latest OpenAI models: GPT-5.X” with “Priority” access across all three plans (Business Premium, Business Standard, and the Copilot Business add-on). The page lists “Model Choice” as a linked feature for the Copilot Business add-on specifically. The page does not publish a per-user token cap or per-prompt context window for Copilot at the plan level, and Microsoft 365 Copilot for individuals is documented on the individual product page rather than the business pricing page.

The Organisation tier also adds “Work IQ” grounding, “Pre-built Microsoft agents, including Researcher, Analyst, and Facilitator,” and “Copilot in Word, Excel, PowerPoint, Outlook, and Microsoft Teams.” Copilot Chat is described as “a web-grounded AI assistant available, at no additional cost, with eligible Microsoft 365 plans,” which is the free-with-Enterprise-licence path.

What each plan adds over the free tier

The free tier is the baseline. Here is what the paid plan adds, in each vendor’s own words.

PlanAdds (from the pricing page)
Claude Pro”Claude Code,” “Claude Science,” “Claude in Chrome and Microsoft 365,” “Research,” “Hand off and schedule tasks,” “Claude Dashboards (beta),” “Projects”; model access to Opus and Fable via usage credits; “at least 5x more usage per 5-hour session than Free”
Google AI Pro”4x higher usage limits than free”; “Access to Gemini 3 Pro model in Google Search”; “1,000 Google Flow credits”; “5x more audio summaries”; “Google Home Premium (standard)”; “YouTube Premium Lite”; “5 TB storage”
Microsoft 365 Copilot (org)Priority access to “Latest OpenAI models: GPT-5.X”; “Model Choice”; “Work IQ” grounding; “Pre-built Microsoft agents, including Researcher, Analyst, and Facilitator”; “Copilot in Word, Excel, PowerPoint, Outlook, and Microsoft Teams”; one-month free trial then auto-conversion to a paid subscription

Three patterns stand out. First, the model tier you can reach is the single biggest differentiator: Claude Pro unlocks Opus and Fable; Google AI Pro unlocks the Gemini 3 Pro Search path; Microsoft 365 Copilot’s differentiation is story-driven against the base OpenAI access rather than a model-tier jump. Second, the integration surface is the second biggest differentiator: Claude Pro bolts onto Chrome and Microsoft 365; Google AI Pro comes with Google Home Premium and YouTube Premium Lite; Microsoft 365 Copilot lives inside Word, Excel, PowerPoint, Outlook, and Teams. Third, the usage cap is stated differently everywhere: Claude Pro says “at least 5x more usage per 5-hour session than Free,” Google AI Pro says “4x higher usage limits than free,” and Microsoft 365 Copilot says “Priority” without a published token cap.

Privacy and training posture

Each vendor states its own training policy on the consumer chat products; that policy is what determines whether your chat history is fair game for future model training.

  • Claude is the most conservative on this point. Anthropic’s consumer policy is to not train on Free or Pro chat inputs by default; the converse, training on prompts, requires an explicit opt-in. The Claude pricing page does not state the training policy itself; the policy page is the load-bearing reference.
  • Google AI Pro inherits Google’s policy of not training on paid Gemini app and Gemini API inputs. Workspace customers get enterprise data handling on Gemini for Workspace by default; consumer AI Pro is a separate but adjacent posture. The free-tier policy is more permissive and worth checking before you commit.
  • Microsoft 365 Copilot inherits Microsoft 365’s enterprise data protection on eligible plans; the page explicitly states “Copilot Chat is a web-grounded AI assistant available, at no additional cost, with eligible Microsoft 365 plans.” Consumer Microsoft 365 Premium inherits Microsoft’s standard consumer-data policy, which is more permissive than the Enterprise tier.

None of these vendor-published pricing pages is the source-of-truth for the training policy. Check the vendor’s privacy or data usage page before relying on the chat product for sensitive inputs.

Picking by use case

A simple decision tree covers most readers.

  1. Pick Claude Pro if your main job is long-document work and clean chat. Sonnet 5.5 and Opus 5.5 on Pro both run in the same “Up to 1M” context window that Free offers, but Pro raises the per-session rate limit to “at least 5x more usage per 5-hour session than Free.” Pro adds Claude Code and Projects, which are the practical picks for code and long-document workflows.
  2. Pick Google AI Pro if your work already lives in Google Docs, Sheets, and Gmail. AI Pro unlocks the Gemini 3 Pro path through Search, raises the usage cap to “4x higher usage limits than free,” and bundles Google Home Premium (standard), YouTube Premium Lite, and 5 TB of Google One storage. AI Pro is the consumer tier that bundles the AI features with the Google One storage plan.
  3. Pick Microsoft 365 Copilot if your work lives in Word, Excel, Outlook, and Teams. The Copilot Business add-on starts at $18/user/mo annual on top of a qualifying M365 licence. The plan adds “Latest OpenAI models: GPT-5.X” with priority access, “Model Choice,” “Work IQ” grounding, and the Researcher, Analyst, and Facilitator agents. The one-month free trial converts to a paid subscription automatically, so set a reminder to cancel if the trial is the value.

What to skip

Three things are worth calling out.

  1. Assuming the free tier is enough. The free tier on all three vendors throttles heavy users. Claude Free limits model choice to Sonnet and Haiku and caps usage per 5-hour session; Google AI Free limits Gemini app usage and offers only “limited access to 3.1 Pro”; Microsoft 365 Copilot Chat is “$0” but is web-grounded and metered. If you are hitting any of these ceilings, the paid tier is the difference between getting blocked mid-task and not.
  2. Paying twice for the same model. Microsoft 365 Copilot and the ChatGPT product both run on OpenAI models; if you already pay for a ChatGPT subscription and rarely use Microsoft 365, the Office integration is not worth the marginal cost. Conversely, if your employer already covers Microsoft 365 Copilot, paying for a separate ChatGPT subscription duplicates model access.
  3. Trusting the model name without checking the tier. Google AI Pro unlocks “Gemini 3 Pro model in Google Search” but the actual Gemini app still rate-limits “Gemini 3.1 Pro” usage; Claude Pro includes Opus but caps “50% of weekly limits” for Max users and uses usage credits for Fable; Microsoft 365 Copilot advertises GPT-5.X with priority access but does not publish a per-user token cap. Read the model’s specific quota, not the marketing line, before paying.

The bottom line

The $20/mo floor is a real tier with three serious contenders. Claude Pro is the right pick for clean, long-context chat with Opus and Fable available via usage credits. Google AI Pro is the right pick if you want Gemini 3 Pro through Search and Google One storage bundled in. Microsoft 365 Copilot is the right pick if you live in Office and want priority access to GPT-5.X with Researcher, Analyst, and Facilitator agents. Each vendor publishes its pricing and feature list on the same surface as the rate limits; the URLs above are the load-bearing references and are worth re-checking before you commit, because every vendor changed a price or model tier in the last 12 months and will change them again.