Start here

Context window vs usage window

A context window limits what a model can see at once; a usage window limits how much you can use in a period. Confusing them causes most limit surprises.

Quick answer

A context window is the maximum text a model can consider in one request: instructions, conversation, files, and output. A usage window is the allowance you can consume over a period, such as five hours or a week. Filling the context raises the cost of each turn; exhausting the usage window blocks you entirely.

Side by side

Context windowUsage window
LimitsOne requestA period of work
RefillsStart a new sessionWhen the period resets
SymptomTruncation or summarisationRequests blocked
Grows withFiles and historyModel calls over time
FixSmaller scopeWait, switch model, or upgrade

The common mistake

People see a percentage next to a model name and assume it is their plan allowance. Usually it is context. Clearing the conversation resets one and does nothing to the other, which is why 'I restarted and it still says I am limited' is such a common report.

Read the label: context percentages usually sit next to the model or the window; plan usage sits under an account or usage heading.

What to do about each

  • Context tight: start a new session, point at fewer files, summarise.
  • Usage tight: switch to a cheaper model, wait for the reset, or use another tool.
  • Both tight: reduce scope first, then decide about the plan.

The Codex context guide covers the first, and the glossary defines the rest.

Questions, answered

Is the context window the same as my usage limit?

No. Context limits a single request; usage limits a period. Clearing a conversation resets context but not plan usage.

Does a bigger context window cost more?

Yes, because every turn sends the context. A larger window does not mean larger spend; using more of it does.

Why am I blocked with context available?

Your usage window is exhausted. Those are separate counters.

Browse by topic