How big is ChatGPT's context window?

It depends on the model and the plan, and it changes as OpenAI releases new models. OpenAI lists the context window for each model in its documentation. The chat app can also use less than the model's full window.

Is a context window the same as AI memory?

No. The context window is the working space for one response. Memory features, like ChatGPT's saved memories, store facts between conversations and add them back into the context window when they're relevant.

Does a bigger context window reduce hallucinations?

Not by itself. A bigger window lets a model see more material, but research shows models use information in the middle of long inputs less reliably. What helps is putting the right material in the window, which is what retrieval and grounding do.

What Is a Context Window? How Much AI Can Read at Once

Definition

Context windowA context window is the maximum amount of text an AI model can work with at once, measured in tokens. It holds everything in play for a single response: the instructions, the conversation so far, any web pages or documents the system retrieved, and the model's own reply. Anything that doesn't fit is left out.

How Searchable measures this

When an AI engine searches, only the passages it retrieves reach its context window. Searchable's page audits check the structure that helps a passage stand on its own, such as headings, an answer-first summary, and an FAQ section, and fail pages with almost no readable text.

How a context window works

Everything a model considers for one response has to fit in its context window, and the answer the model writes counts toward the limit too. When a conversation outgrows the window, older material gets dropped or summarized.

The window is measured in tokens, not words. Anthropic's documentation gives a rough conversion: 200,000 tokens is about 150,000 words of English.

How big context windows are

Context windows have grown quickly. Anthropic's current Claude models list a 1 million token window, which is several novels' worth of text. Other vendors publish their own limits per model, and the numbers change with each release.

The advertised size is the most a model can accept. A consumer chat app may use a smaller window than the model supports, and some vendors charge more for very long inputs.

Bigger context windows don't mean better reading

A model can accept a long input without using all of it equally well. In the 2023 paper Lost in the Middle, researchers found that models performed best when the relevant information sat at the beginning or end of the input, and worse when it sat in the middle.

Newer models handle long inputs better, but the practical point holds. Putting more text in front of a model doesn't guarantee it'll use the part that matters.

How context windows work in AI search

AI search engines don't load whole websites into the context window. They run searches, retrieve the passages that look most relevant, and pass those to the model with your question. Each passage competes for space with every other source.

The competition has grown as engines read more. In Searchable's research depth study, the number of pages ChatGPT showed it had consulted per answer went from 11.6 to 22.5 overnight on August 8, 2026. More pages in the window means each one gets a smaller share of the model's attention.

What context windows mean for your content

Since engines retrieve passages rather than pages, each section of a page has to make sense alone. A few habits help:

Put the direct answer near the top of the page and at the start of each section.
Keep each section about one thing, so a retrieved chunk carries a complete point.
Name the subject in the passage itself instead of relying on "it" or "this tool" from earlier.

Content chunking covers how pages get split into these passages.

Frequently asked questions

A token is a chunk of text a model reads as one unit, often a whole short word or part of a longer one. Token counts vary by model and language, so the same text can use a different number of tokens in two models.

See how AI engines talk about your brand

Get a free report on where ChatGPT, Perplexity, and Google AI name and cite you.

Get your free visibility report

TokenizationBreaking text into tokens, the small units of words and characters that language models read and generate.Content chunkingContent chunking is the way AI search systems split a web page into smaller passages so each one can be retrieved and quoted on its own. The term also describes writing pages in clear, self-contained sections, so any passage an engine pulls out still makes sense without the rest of the page.Retrieval-augmented generation (RAG)Retrieval-augmented generation (RAG) is a technique where an AI model first searches a set of documents for passages that match a question, then writes its answer from those passages. Because it looks things up at answer time, the model can use up-to-date information and point to where it came from.

Large language model (LLM)An AI model trained on large amounts of text to predict and generate language. LLMs are the engines inside ChatGPT, Claude, Gemini, and most AI search tools.

Agentic searchAgentic search is when an AI system runs its own research loop instead of a single search. It plans queries, reads the results, decides what's still missing, and searches again until it can answer. Deep research modes in ChatGPT, Gemini, Perplexity, and Claude work this way, and they read far more pages than one search returns.Vector searchVector search finds content by meaning instead of exact words. It turns text into embeddings, lists of numbers that represent what the text means, then returns the stored items whose embeddings sit closest to the query's. That's how a search for "cheap flights" can find a page about "budget airfare."

Browse the full AI search glossary