Does Claude's free plan have ads?

No. Anthropic said on 4 February 2026 that "Claude will remain ad-free", and that covers the free plan. Claude Free runs Sonnet and Haiku, with usage that resets on a rolling five-hour window.

What is the cheapest paid AI subscription?

In the US, Google AI Plus at $4.99 a month is the cheapest plan from a major AI company, followed by $5 apps such as Plainly and ChatGPT Go at $8. Google AI Plus doubles the Gemini free usage limits and adds 400 GB of storage; Go may show ads.

Is Gemini free to use?

Yes. The Gemini app is free with a Google account and runs Gemini 3.5 Flash-Lite, with Deep Research and image generation at lower limits. Paid plans start at $4.99 a month for Google AI Plus and $19.99 for Google AI Pro, which adds a 1 million token context window.

What happens when you hit Claude Pro's usage limit?

You wait for the five-hour session to reset, turn on usage credits billed at standard API rates, or move up to Max at $100 or $200 a month. Weekly limits apply on top of the session limit, and Claude Code draws on the same allowance.

Is ChatGPT Go better than the free plan?

A little. Go costs $8 and runs the same models as Free, still with ads, but doubles the Instant context window from 27K to 54K tokens and adds scheduled tasks and more image generation. Its reasoning mode reaches 256K tokens, the same as Plus.

Free vs paid AI apps: every tier compared

ChatGPT, Claude, Gemini, Copilot and Perplexity free vs paid, from $4.99 to $500 a month, and why heavy users should skip the $20 plan.

Free vs paid AI apps: what you get when you pay

Free AI apps give you a capable model with limits on usage, context or features. Paying buys bigger models, more room per chat, higher limits and extras such as deep research and video. US plans run from $4.99 to $500 a month, and the right one depends on how much you use AI.

The usual advice is to start free and upgrade as you use more. My rule runs the other way for the middle of the ladder, and the reason is usage caps.

Which plan to buy for how much you use AI

How much you use AI	My pick	Price	Why
A little	Claude Pro	$20 ($17 billed yearly)	Opus, Sonnet and Haiku with no ads, and light use stays under the limits
A lot	Plainly	$5	Unlimited chat on a capable open-source model, so there is no cap to run into
A huge amount	Claude Max 5× or ChatGPT Pro	$100	A great amount of usage, right for everything except the most demanding programming, at a lot of money

I built Plainly, so weigh the middle row with that in mind.

Why heavy users should skip the $20 plan

A $20 frontier plan only makes sense if you never hit its usage cap. Light users won't. They get Opus and Sonnet for $20 and never meet the ceiling. Heavy users hit it in the middle of work, and then they wait for the limit to reset, pay extra, or move up to $100.

Claude Pro shows the mechanic clearly. Anthropic gives Pro "at least 5x" the usage of Free per five-hour session, adds weekly limits on top, and counts claude.ai, the desktop app and Claude Code against the same pool (claude.com/pricing and Anthropic's help center, October 2026). When you run out, you can turn on usage credits at standard API rates or upgrade.

I was on Claude Pro, billed at US$22 a month, for a long time, and I always ran out. So did my friends. Around January 2026 I moved to Max 5× at US$100 a month, and I still hit its limit. For someone who asks a lot of everyday questions, a $5 plan with unlimited chat on a model slightly behind the frontier beats a $20 plan that stops them until the session resets or charges extra. For someone who works with AI all day, the answer is the $100 plan.

ChatGPT works differently. Every ChatGPT plan, free included, has unlimited everyday text chats, subject to OpenAI's abuse guardrails. What changes as you pay is the model, the context window and the tools. For ChatGPT Plus specifically, my rule is to pay only if you program, your work pays for it, or you use it professionally. Otherwise you don't need to pay.

Free AI apps compared

Free plan	Model you get	Limits	Extras on the free plan
ChatGPT	GPT-6 Luna, GPT-5 Thinking Mini	Unlimited everyday text chats; 27K-token Instant context (about 12 pages)	Limited deep research, image generation, memory and Codex
Claude	Sonnet and Haiku	Usage limited, resets on a rolling five-hour window	Covers everyday questions; no Opus or Fable
Gemini	Gemini 3.5 Flash-Lite	Lower usage limits than paid plans; Google publishes no fixed count	Deep Research, image generation and editing
Microsoft Copilot	Copilot's standard models	Free; a Microsoft account is optional	Chat and web answers; more Copilot use comes with Microsoft 365 plans
Perplexity	Basic AI models	Limited use; no fixed count published	Cited web answers; the model picker and higher limits need Pro

Sources: chatgpt.com/pricing and claude.com/pricing (11 October 2026); Google's US Gemini plans page (October 2026); Microsoft's Copilot privacy FAQ and Microsoft 365 plan page (October 2026); Perplexity's pricing page (18 October 2026).

Ads differ across the free plans. ChatGPT Free and Go may show ads, which OpenAI began testing in the US on 9 February 2026. Anthropic pledged on 4 February 2026 that "Claude will remain ad-free." Google says Gemini app chats are not used to show you ads, and that it will tell users if that changes (Gemini privacy hub, September 2026). Perplexity launched sponsored follow-up questions in November 2024 and has since wound the program down, according to the Financial Times.

I think ChatGPT's free plan is good, and you can't complain about free. Its limit is room: at 27K tokens of Instant context, a long chat loses its early details sooner than on any paid plan. Claude Free runs Sonnet with no ads, and its catch is the session limit.

Paid AI plans compared, from $4.99 to $500

Plan	US price per month	What paying adds	Ads	Catch
Google AI Plus	$4.99	2× Gemini free usage, 128K context, video generation, 400 GB storage	Chats not used for ads, per Google	Gemini models only
Plainly	$5	Unlimited chat, web search with sources, documents and photos	No	No image generation, voice, memory across chats or phone app
ChatGPT Go	$8	54K Instant and 256K reasoning context, scheduled tasks, more images	May show	Same models as Free
Claude Pro	$20 ($17 yearly)	Opus, Sonnet and Haiku; at least 5× Free usage per session	No	Five-hour and weekly limits; Fable only through usage credits
ChatGPT Plus	$20	GPT-6.1 Sol, GPT-6 Astra, deep research, Codex	No	Instant context stays at 54K, same as Go
Google AI Pro	$19.99	Gemini's Pro model at higher limits, 1M-token context, 5 TB storage	Chats not used for ads, per Google	Gemini models only
Perplexity Pro	$20	More Pro searches, Deep Research and the model picker	No	Built around search
Microsoft 365 Premium	$19.99	The most Copilot use, the Office apps and 2 TB of OneDrive storage	Not stated by Microsoft	Best value if you already use Office
Claude Max	$100 (5×) or $200 (20×)	5× or 20× Pro usage; Fable up to half of weekly limits	No	Still capped
ChatGPT Pro	$100, $200 or $500	128K Instant context, Pro reasoning on GPT-6 Astra; Ultrafast on eligible tiers	No	Allowances differ by tier
Google AI Ultra	$99.99 or $199.99	5× or 20× AI Pro usage, Deep Think	Chats not used for ads, per Google	Gemini models only
Perplexity Max	$200	Highest limits, Perplexity Computer credits	No	Price

Sources: claude.com/pricing and chatgpt.com/pricing (11 October 2026); Google's US Gemini plans page (11 October 2026) and 9to5Google on AI Plus's 128K context (8 June 2026); Microsoft's Microsoft 365 plan page (October 2026); Perplexity's US App Store listing for its $20 and $200 prices. Plans change often, so check the vendor's page before you buy.

What paying for AI adds

The first thing a paid plan buys is a bigger model. ChatGPT Plus moves you from GPT-6 Luna to the GPT-6 Sol family, Claude Pro adds Opus, and Google AI Pro raises your access to Gemini's Pro model. On everyday writing the gap is small: in a five-task writing test on 2 September 2026, ChatGPT on Go and Plainly each scored 61 of 75 and Claude's Fable 5 scored 59. Programming is the first reason on my list for paying for Plus.

Paying also buys room. Context windows grow from 27K tokens on ChatGPT Free to 54K on Go and Plus in Instant mode, 256K in reasoning mode, and 1 million tokens on Google AI Pro. A bigger window means the model keeps hold of a long document or a long chat.

Higher plans then add volume and tools: bigger usage allowances, deep research, video generation, coding agents and more image generation, mostly from $20 up.

The cheap tiers: $4.99, $5 and $8

Google AI Plus is the cheapest paid plan from a major lab. My view of it: good, but you're tied to Google's models, and Google's models aren't currently state of the art.

Plainly costs $5 a month with unlimited chat. Plainly runs one capable open-source model and switches when a better one comes out. You never pick. It searches the web when a question needs current information and links its sources, reads documents and photos, and syncs chats across devices in the browser. It does not generate images or video, has no voice mode or phone app, and keeps no memory between chats. New accounts can ask a limited amount free each month before paying.

ChatGPT Go at $8 is a bad tier. It still shows ads, it runs the same smaller models as Free, and its default Instant mode holds 54K tokens. I have Go through my Revolut subscription, and on the default Instant model it lost things I had told it earlier in the same chat, mid-task. Plus has the same 54K Instant window, so upgrading from Go only fixes that if you switch to reasoning mode.

The $20 tier

For a ranking of the apps themselves rather than the price tiers, see the best AI apps.

Claude Pro is the plan I'd pick for light use, and the one most likely to run out on a heavy user, because of its five-hour and weekly usage limits.

ChatGPT Plus and Google AI Pro buy the top models from OpenAI and Google, and Google bundles 5 TB of storage that some people would pay for anyway. Microsoft 365 Premium makes most sense if you already pay for Office.

I had a free year of Perplexity Pro through my phone company, Optus. I used it heavily for a few weeks, then not at all for ten months, and I was glad I hadn't paid. My impression, which is anecdotal, was that answers felt weaker than the same models used directly, and I don't like the interface.

Every main plan sits within a cent of $20, the price ChatGPT Plus launched at in February 2023, and OpenAI still loses money at that level.

Where this comparison falls short

The Gemini, Copilot and Perplexity rows come from the companies' published plans, and I've used Gemini through its website only, never its app. Model names change every few months, so the model columns will date first. The prices were checked on 11 October 2026.