Is the biggest context window automatically the best chatbot?

No — window size only says how much text fits, not how well the model suits your work. Gemini 1.5 Pro leads at 1,000,000 tokens, but if your largest job is a 2,000-word draft (about 2,700 tokens), even GPT-4o's 128,000-token window is more than forty times bigger than the task.

Can I paste a whole book into one chat?

Yes, if you pick the right window. A 500-page book is about 200,000 tokens: it fills one Claude Sonnet window, spills across 1.6 GPT-4o windows, and fits inside Gemini 1.5 Pro's million-token window with room to spare.

Why does one person pay $20 a month while another pays per token?

They are two different doors into the same kind of model. ChatGPT Plus and Claude Pro are flat subscriptions at $20 per month for one user; API access is metered at per-million-token rates, such as $2.50 per million input tokens on GPT-4o.

How can I estimate my cost before choosing?

Count words, multiply by 1.3, then apply the listed rates. A 100,000-token input plus a 20,000-token output at $2.50 and $10 per million is $0.45 per call — so a $20 flat month equals roughly 44 such calls.

1M tokensGemini 1.5 Pro's context window — the biggest here

$20 / monthChatGPT Plus or Claude Pro, one individual user

$0.45a 100K-in, 20K-out GPT-4o API call

1,000,000tokens in the largest window (Gemini 1.5 Pro)

The biggest window, priced

AI Fervour — AI chatbots and assistant models compared | aifervour.com

Gemini 1.5 Pro holds the biggest context window of the major chatbots: 1,000,000 tokens, roughly 750,000 English words in a single pass. Claude Sonnet follows at 200,000 tokens and GPT-4o at 128,000. As for the monthly cost, the two flagship subscriptions — ChatGPT Plus and Claude Pro — each list at $20 for one individual user, while API access is metered per million tokens.

aifervour.com
AI Fervour — an independent notebook on context windows and token costs.
Context windows and token prices, converted into words and dollars.

Numeric index

The question, answered in numbers

1,000,000tokens in the largest window (Gemini 1.5 Pro)Converted at 0.75 words per token, that is about 750,000 English words.

200,000tokens in a 500-page bookIt fills one Claude Sonnet window but needs 1.6 GPT-4o windows.

$0.027cost of a 2,000-word draftAbout 2,700 output tokens at $10 per million — under three cents.

Windows and prices in one table

Context windows and listed prices across the major chatbots and API tiers.
Model or plan	Window or listed rate	What that means in practice
Gemini 1.5 Pro	1,000,000 tokens	about 750,000 words — the biggest window here
Claude Sonnet	200,000 tokens	about 150,000 words, roughly a 500-page book
GPT-4o	128,000 tokens	about 96,000 words; a 200,000-token job needs 1.6 windows
ChatGPT Plus	$20 per month	flat subscription for one individual user
Claude Pro	$20 per month	flat subscription for one individual user
GPT-4o API input	$2.50 per million tokens	a 100,000-token prompt costs $0.25
GPT-4o-mini API input	$0.15 per million tokens	about one-seventeenth of the GPT-4o input rate
GPT-4o API output	$10 per million tokens	a 2,000-word draft (2,700 tokens) costs under $0.03

Context windows and listed prices across the major chatbots and API tiers.

Which chatbot has the biggest context window?

Gemini 1.5 Pro — one million tokens, about 750,000 words in one window.

Gemini 1.5 Pro accepts the most text at once: a 1,000,000-token window, roughly 750,000 English words. Claude Sonnet holds 200,000 tokens (about 150,000 words) and GPT-4o 128,000 (about 96,000).

In book terms, a 500-page book is about 200,000 tokens: it fills one Claude window, spills across 1.6 GPT-4o windows, and occupies only a fifth of the Gemini window.

Biggest is not automatically best. If your largest paste is a 2,000-word draft — about 2,700 tokens — even the smallest of these windows is more than forty times larger than the job.

Terms this page relies on

Token: The unit every model reads and bills. One English token averages about four characters, or 0.75 words, so 1,000 words come to roughly 1,300 tokens.
Context window: The most tokens a model can hold at once — your text plus its answer. GPT-4o's is 128,000; Gemini 1.5 Pro's is 1,000,000.
Input and output tokens: What you send versus what the model writes back, billed at separate rates: $2.50 per million in, $10 per million out on GPT-4o.
Temperature: An API dial from 0 to 2. Zero pins output to the most likely continuation; two maximises sampling spread. It changes wording, not the price.
Flat subscription: A fixed monthly price for one individual user — $20 for ChatGPT Plus or Claude Pro — instead of per-token billing.

Objections readers actually raise

Is the biggest context window automatically the best chatbot?: No — window size only says how much text fits, not how well the model suits your work. Gemini 1.5 Pro leads at 1,000,000 tokens, but if your largest job is a 2,000-word draft (about 2,700 tokens), even GPT-4o's 128,000-token window is more than forty times bigger than the task.
Can I paste a whole book into one chat?: Yes, if you pick the right window. A 500-page book is about 200,000 tokens: it fills one Claude Sonnet window, spills across 1.6 GPT-4o windows, and fits inside Gemini 1.5 Pro's million-token window with room to spare.
Why does one person pay $20 a month while another pays per token?: They are two different doors into the same kind of model. ChatGPT Plus and Claude Pro are flat subscriptions at $20 per month for one user; API access is metered at per-million-token rates, such as $2.50 per million input tokens on GPT-4o.
How can I estimate my cost before choosing?: Count words, multiply by 1.3, then apply the listed rates. A 100,000-token input plus a 20,000-token output at $2.50 and $10 per million is $0.45 per call — so a $20 flat month equals roughly 44 such calls.

How this page was compiled

Compiled from published model specifications and listed prices: the stated windows for GPT-4o, Claude Sonnet and Gemini 1.5 Pro, and the $20 monthly listings for ChatGPT Plus and Claude Pro. Word equivalents use the published 0.75-words-per-token average; costs are worked at the listed $2.50 and $10 per-million-token rates. Where a figure is not published, the page says so instead of estimating.

Where the numbers come from

Every number traces to the providers' own specification sheets and price lists.