Buyer's question

What Is a Context Window? The amount of text an LLM can read at once.

Context window is a term that comes up constantly in AI buying decisions. Here's a plain-English explainer of what it is, why it matters, and what 'long context' actually unlocks.

Short answer

A context window is the maximum amount of text (and other data) an LLM can read and reason about in a single conversation. Measured in tokens (roughly 4 characters or 0.75 words). Claude Sonnet 4.7 supports 200K tokens standard, with 1M available on some tiers; GPT-5 class is similar. 1M tokens is about 750,000 words — a 2000-page book.

Why context window matters

Longer context window means the model can hold more in its 'attention' at once. Practical implications:

Big documents: Analyze 200-page PDFs without splitting.
Long conversations: Hold extensive history without forgetting earlier turns.
Multi-document synthesis: Compare and reason across 20 documents at once.
Large codebases: Work with many files of context (especially valuable for Claude Code).
Loaded knowledge: Projects with extensive knowledge work better with longer context.

What 'long context' unlocks practically

Load all relevant past contracts into a contract drafting Project; the model has them all in mind.
Paste a full RFP document for synthesis instead of summarizing it first.
Have a 2-hour transcribed meeting analyzed end-to-end.
Code refactors that span dozens of files.

Limits to be aware of

Performance degrades at extreme lengths. Models can struggle to attend to information in the middle of very long contexts.
Cost scales with tokens. Bigger context = more tokens = more cost.
Latency increases. Longer prompts take longer to process.
Quality of output may not improve. Stuffing more context isn't always better; relevance is.

Practical guidance

Most B2B workflows do not need extreme context. 200K tokens (Claude Sonnet standard) handles 99% of business tasks easily. Reach for 1M context only when you have a specific reason — analyzing very long documents, full codebase reasoning.

FAQ

How big is a token?

About 4 characters or 0.75 words on average. 1,000 tokens is roughly 750 words.

Is bigger context always better?

No. Bigger is better when needed; otherwise wastes cost and can slow output. Right-size for the task.

What's Claude's context window?

Claude Sonnet 4.7 supports 200K tokens standard; 1M tokens available on certain tiers.

What's GPT-5's context window?

Comparable to Claude — large, variable by tier.

Should context window be a primary buying criterion?

Not by itself. Quality, integration, and workflow fit matter more for most teams.

Want a roadmap built for your business?

The $1,500 AI Audit produces a written, prioritized roadmap in 5 business days.

Book the AI Audit → Take the Gap Assessment