ChatGPT Plus, Claude Pro and Google AI Pro all cost about $20 a month, and all three will happily answer your emails, summarize a PDF and write a Python script. The differences are in what you get past the basics: which models, how much you can use them before a cap kicks in, which agents and coding tools are bundled, and what extras come along for the ride. This comparison lays out the five criteria that matter, puts the plans side by side, and ends with a recommendation for each kind of user. All prices and limits are as of August 2026.
Which $20 plan to pick: Claude Pro, Google AI Pro or ChatGPT Plus
- Best for writing and long documents: Claude Pro. Opus 5's 1M-token context and Anthropic's prose quality are hard to beat, and the plan now includes Claude Code, Cowork and Design.
- Best all-rounder for people who already live in Gmail, Docs and YouTube: Google AI Pro. Same price, plus 5 TB of storage, Gemini Notebook, Veo video and YouTube Premium Lite.
- Best for the widest feature set and the biggest ecosystem: ChatGPT Plus. GPT-5.6 Sol reasoning, Deep Research, browser agents and the largest library of integrations.
If you can only pick one and have no strong pull toward Google's ecosystem, Claude Pro or ChatGPT Plus is a coin flip decided by taste: Claude for careful writing and coding, ChatGPT for breadth.
How we compared them: the five criteria
Three plans, five questions:
- Price and billing. What you pay monthly, whether an annual discount exists, and what sits above and below the tier.
- Models. Which models you get and whether you get them at full strength.
- Usage limits. How much you can actually use before the plan throttles you, and how the window is measured.
- Agents, coding and research tools. What the plan includes beyond chat.
- Extras. Storage, video, media perks and whether you see ads.
Here is the summary table; the sections that follow go criterion by criterion.
Scroll sideways to see the full table →
| ChatGPT Plus | Claude Pro | Google AI Pro | |
|---|---|---|---|
| Monthly price | $20 | $20 | $19.99 |
| Annual price | No discount | $17/month ($200/year) | No standard discount |
| Flagship model | GPT-5.6 Sol (medium and high effort) | Claude Opus 5 | Gemini 3.1 Pro |
| Context window | Varies by model | 1M tokens on Opus 5 | 1M tokens |
| Limit style | Unlimited plain text since August 2026; separate caps on thinking, images, files and Deep Research | Rolling 5-hour session plus weekly cap | Compute budget refreshing every 5 hours, weekly cap |
| Coding tool | Codex | Claude Code (shared usage) | Gemini CLI, Antigravity (limited) |
| Research agent | Deep Research (monthly allowance) | Research in Claude | Deep Research |
| Storage | None | None | 5 TB |
| Video generation | No consumer app since April 2026 | No | Veo |
| Ads | None on Plus | None | None |
| Cheaper tier | Go, $8 | None | Plus, $7.99 |
| Pricier tier | Pro, $100 / $200 | Max, $100 / $200 | Ultra, $99.99 / $199.99 |
Price and billing: $20, $19.99 and the annual discount
At list price the three are effectively identical. ChatGPT Plus is $20 a month, Claude Pro is $20, Google AI Pro is $19.99.
The gap opens on annual billing. Claude Pro drops to $17 a month when billed annually, $200 up front. Neither OpenAI nor Google offers a standing annual discount on the $20 tier, though both run promotional first-month offers from time to time.
There is also a difference in what one subscription covers. Claude Pro is a single subscription across Claude's chat apps, Claude Code, Claude Cowork, Claude Design and Claude Science, all drawing from one usage pool. ChatGPT Plus covers ChatGPT and Codex. Google AI Pro covers the Gemini app, Gemini in Workspace apps, Gemini Notebook and the storage and YouTube perks.
Which models you actually get on each plan
This is where the plans diverge most, because each company ships a top model and a cheaper model and gates them differently.
ChatGPT. OpenAI released the GPT-5.6 family on July 9, 2026 as three variants: Luna (the cheapest), Terra (the balanced model) and Sol (the flagship for hard problems). Plus, Pro, Business and Enterprise users get GPT-5.6 Sol at medium and higher effort settings in the model picker. Free and Go users get GPT-5.6 Luna as their default, with a Think button that escalates a single hard question. Plus is the cheapest tier at which you get the actual frontier model.
Claude. Anthropic released Claude Opus 5 on July 24, 2026. It has a 1M-token context window, up to 128K tokens of output, adaptive thinking on by default and an effort control that runs from low through max. Opus 5 is the default model on Claude Max and the strongest model available on Claude Pro. Pro users get the same model as Max users; they just get less of it.
Gemini. Google AI Pro includes what Google calls the full model lineup, meaning Gemini 3.1 Pro (released February 19, 2026) at higher limits than the free and Plus tiers, plus the Flash models for everyday work. Gemini 3.1 Pro is also the model behind Gemini Notebook for Pro subscribers.
The practical upshot: on all three plans you get the company's best model. None of them hold the flagship back for the $100-plus tiers anymore; those tiers sell volume, not access.
Usage limits: messages, compute windows and weekly caps
Limits are the thing people notice after the first week, and the three companies measure them in three different ways.
ChatGPT Plus is the most legible. For most of 2026 OpenAI's published cap was about 160 messages every three hours on the default model, on a rolling window. In early August it made plain text chat unlimited on every tier, including Free, so the limits that remain are the separate allowances for thinking-mode messages, image generation, file uploads and Deep Research, which OpenAI adjusts with demand and does not always publish.
Claude Pro uses two windows: a rolling five-hour session limit and a weekly cap that applies across all models. Anthropic does not publish exact message counts, because what counts is tokens, and a 200-page PDF in context costs far more than a one-line question. The important point is that Claude Code, Cowork and chat all draw from the same pool, so a heavy coding afternoon eats into your chat allowance.
Google AI Pro moved to a compute-based system in 2026: you have a budget that refreshes every five hours until you reach a weekly limit, and expensive actions (extended thinking, video generation) burn through it faster than plain chat. Google does not publish the numbers either.
For a light user, none of the three limits will ever appear. For a heavy user, the honest ranking is: ChatGPT Plus is the most predictable, Claude Pro is the most generous per dollar for long-context work but the easiest to exhaust with coding, and Google AI Pro sits in between with the added annoyance that media generation shares the same budget. Check the limits table again in a few months; it is the part of this comparison that changes most often.
Agents, coding and research tools included
ChatGPT Plus bundles Deep Research (with a monthly allowance on Plus), Codex for coding, and the browser-agent features that OpenAI folded back into ChatGPT after it shut down the Atlas browser on August 9, 2026. Atlas lasted less than a year; its capabilities now live in a ChatGPT browser extension and the desktop app rather than a standalone browser. If you want to understand what these agent features are actually doing under the hood, see our explainer on AI agents and MCP.
Claude Pro includes Claude Code (terminal and IDE coding agent), Claude Cowork (desktop agent for files and apps), Claude Design and Claude Science, all on the same subscription. For developers this is the strongest bundle at $20; the catch, again, is that Claude Code shares your usage pool.
Google AI Pro includes Deep Research, Gemini CLI, and access to Antigravity, Google's agentic coding platform, at lower priority than Ultra. Gemini's agent features are less mature than the other two for autonomous tasks, but the Workspace integration (Gemini inside Gmail, Docs, Sheets and Meet) is something neither rival can match.
Extras: storage, video, YouTube Premium Lite, ads
Google wins this section outright. Google AI Pro includes 5 TB of cloud storage, raised from 2 TB on April 1, 2026 at no price change, plus Veo video generation, YouTube Premium Lite, Google Home Premium and Gemini Notebook's higher tier. If you were going to pay for Google One storage or YouTube anyway, the AI plan is close to free.
ChatGPT Plus and Claude Pro have no storage or media bundle. OpenAI wound down the consumer Sora app in April 2026, so video generation is no longer a Plus selling point.
On ads: since February 2026, ChatGPT shows labeled sponsored results to Free and Go users in the US and a growing list of countries. Plus and above are ad-free. Neither Claude nor Gemini shows ads on any tier at the time of writing, though Google's search-adjacent products are an obvious place for that to change.
Best for writing, best for coding, best for Google users
Writers, editors, researchers: Claude Pro. Opus 5 handles very long documents in one go, its prose is the least generic of the three, and the Projects feature keeps your style guide and reference material attached. ChatGPT Plus is a close second with better web search built into research. Gemini is fine but tends toward summary-shaped output.
Developers: Claude Pro if you mainly work in a terminal or IDE, because Claude Code is included. ChatGPT Plus if you want Codex and the broader plugin and integration ecosystem. Google AI Pro only if you are already committed to Antigravity. Heavy coders on any plan will hit the $20 tier's limits and should look at the higher tiers below.
People who live in Google Workspace: Google AI Pro, without much debate. The storage alone covers most of the cost, and Gemini inside Gmail and Docs is a daily convenience that a separate chatbot cannot replicate.
Students and light users: consider not paying at all. ChatGPT Free now includes GPT-5.6 Luna with unlimited text chats, Claude's free tier is capable for short sessions, and Gemini Notebook's free limits are generous. Pay when you hit a cap, not before.
Cheaper and pricier alternatives (Go, Max, Ultra)
If $20 is too much or not enough:
Scroll sideways to see the full table →
| Tier | Price (monthly) | What changes |
|---|---|---|
| ChatGPT Go | $8 | More usage than Free, but Luna as the default model and ads in supported countries |
| Google AI Plus | $7.99 | More Gemini usage and 200 GB storage; no Gemini 3.1 Pro in Notebook |
| ChatGPT Pro | $100 (5x) or $200 (20x) | Five or twenty times Plus usage and Codex capacity; the $200 tier adds the highest limits and priority access |
| Claude Max | $100 (5x) or $200 (20x) | Five or twenty times Pro's usage, Opus 5 as default |
| Google AI Ultra | $99.99 (5x) or $199.99 (20x) | 5x or 20x Pro usage, 20 TB storage, full YouTube Premium, Gemini Spark agent; the $199.99 tier adds Project Genie and Mariner |
OpenAI added the $100 Pro tier in April 2026, Google's split of Ultra into two tiers in May 2026 lines up neatly with Claude Max, and all three companies now sell a roughly $100 and a roughly $200 tier. At those prices the question is no longer which company has the best model but which agent you want to run for hours at a time.
Sources and further reading
- 2026 AI Subscription Prices: Gemini vs ChatGPT vs Claude (SentiSight)sentisight.ai
- Claude Subscription Plans & Pricing 2026 (IntuitionLabs)intuitionlabs.ai
- What's new in Claude Opus 5 (Claude Platform Docs)platform.claude.com
- Anthropic releases new model, Opus 5 (Axios)axios.com
- GPT-5.6: Frontier intelligence that scales with your ambition (OpenAI)openai.com
- Improving GPT-5.6 Sol in ChatGPT and expanding access to GPT-5.6 Luna for free users (OpenAI)openai.com
- ChatGPT Plans Compared (Aug 2026): Go vs Plus vs Pro (The AI Career Lab)theaicareerlab.com
- ChatGPT finally offers $100/month Pro plan (TechCrunch)techcrunch.com
- Google's new AI subscription plans explained (TechCabal)techcabal.com
- ChatGPT Now Has Ads for Free and Go Tier Users (MacRumors)macrumors.com
- Evolving Atlas into ChatGPT for browser-based agentic work (OpenAI Help Center)help.openai.com