Clicked Gallery

What is a Context Window?

Highlighted from a real engineering doc. Explained by Clicked.

Used in a sentence

Requests exceeding the model's context window are rejected, so long documents must be chunked before submission.

The reader highlighted one word in the docs. Clicked made the technical term “context window” easy to understand:

Explained in three depths

●○○

Overview

▾

A context window is the maximum amount of text an AI model can consider at once, covering both your input and its reply. Anything beyond that limit is invisible to the model. It is measured in tokens, which are pieces of words rather than whole words.

●●○

Detail

▾

The window covers everything in play: the conversation so far, any documents pasted in, and the answer being generated. Sizes are quoted in tokens, roughly three-quarters of a word each in English, and they vary widely: 128,000 tokens is common, and some models advertise a million, which is several long novels. When a chat outgrows the window, the oldest messages fall out first, which is why a long session forgets instructions given at the start. Bigger windows also cost more to run, since the model compares every token against every other, so doubling the text roughly quadruples that work. Models are weaker in the middle of very long inputs too, recalling the start and end far better, a measured effect known as lost in the middle.

●●●

Analogy

▾

A whiteboard in a meeting room. Everything the team needs must fit on it, and when it fills up, someone erases the oldest corner to make space for new notes. The plan sketched in minute one is gone by hour three, not because anyone chose to forget it, but because the board ran out of surface.

AI explanations may contain errors · Not professional advice

Same facts, different vibe — Slang mode 😎

●○○

Overview

▾

The context window is the AI's working memory: how much text it can hold in its head at once, your words and its own included. Blow past it and the oldest stuff silently falls out. 😎

●●○

Detail

▾

Everything counts against the window: the chat history, that PDF you dumped in, and the reply it's writing. It's measured in tokens, chunks of about three-quarters of a word, and sizes are wild: 128k is standard, a million gets bragged about in launch posts. Overfill it and the oldest messages quietly slide off the cliff, which is why the bot forgets rule one from an hour ago. Bigger isn't free either: the model checks every token against every other, so double the text is roughly four times the work, and your bill feels it. One more gotcha: stuff buried in the middle of a huge dump gets remembered worst, so the killer detail on page 40 of 80 is exactly the one it misses. 😎

●●●

Analogy

▾

Your desk during finals week. Everything you're actually working with has to be on the desk, and when it's full, the oldest pile hits the floor. The formula sheet from Monday isn't gone because you're careless, it's gone because desks have edges.

AI explanations may contain errors · Not professional advice

Formal definition — The same term, explained the usual way

The context window denotes the maximum sequence length, in tokens, that a transformer-based language model can attend to in a single forward pass, comprising both the prompt and generated output. Self-attention scales quadratically with sequence length, constraining practical window sizes, and empirical retrieval accuracy degrades for information positioned mid-sequence.

Want Clicked to explain terms like “context window” directly in your browser — including on PDFs?

Add to Chrome — Free

50 free Explanations · No credit card required

More from the gallery

How to actually read a research paper

Skim smart, break the jargon wall, and keep what you learned — one session, done right.

What is Tokenization?

How AI chops text into chunks before reading — and why some languages cost more.

What is Retrieval-Augmented Generation (RAG)?

Make the AI answer from your documents — retrieval first, generation second, receipts included.

What are Vector Embeddings?

Meaning turned into coordinates — how machines find what's similar without matching a single word.

What does an Idempotent API mean?

Why ten clicks charge you once — the property that makes retries safe.

What is Technical Debt?

Shortcuts and aging systems that bill the future — principal, interest, and repayment.