Context Window Planner
Budget tokens before you run a long prompt.
Runs in your browser. Nothing is uploaded — your input is processed on this device and discarded when you close the tab.
Context Window Planner
How to use this tool
- Enter the model's context window in tokens, from its documentation.
- Add your prompt tokens, system tokens and the output you want to reserve.
- Press Generate for the remaining budget and a rough word estimate.
Worked example
With a 128,000-token window, a 2,000-token prompt, 500 system tokens and a 4,000-token output reserve, 6,500 tokens are used (5.08%) and 121,500 remain, roughly 91,125 words at 0.75 words per token.
What this tool does not do / Limits
Token counts are estimates only: this tool does not count tokens for you, and real tokenisers differ by model, so the same text can count differently elsewhere. The word estimate (0.75 words per token) is a rough English average, not a measurement.
Frequently asked questions
Why reserve output tokens?
Because a reply needs room inside the same window. If the prompt fills the window, the model has nowhere to write its answer and the request fails or is cut short.
How do I find my prompt's token count?
Use the token counting tool provided by your model provider, or paste the text into a tokeniser for that model family; this planner only budgets numbers you supply.
Related tools
JSON Studio
Format, validate, minify, query and compare JSON in your browser, with precise line-and-column errors.
LiveRegex Tester
Test patterns against sample text with match highlighting, capture groups and a safe one-second timeout.
LiveSecurity Header and DNS Checker
Check HSTS, CSP and other security headers, TLS details, and DNS/email authentication records (SPF, DKIM, DMARC).
Live