Clear methods.
Honest limits.

Speed and understanding stay separate

Words per minute = counted words × 60 ÷ reading seconds. For example, 300 words in 90 seconds is 200 words/min. Your pace appears immediately when reading stops. Questions are optional; leaving the check does not count as a wrong answer. Four correct answers out of five is 80% on that question set. It is not a percentage of everything you understood.

What counts as a word?

Our shared counter, english-tokens-v1, counts runs of letters or numbers. It retains internal straight or curly apostrophes and ordinary hyphens: “don’t” and “well-known” each count as one. Spaces, tabs, and newlines separate words; punctuation alone does not count. An em dash splits words. Decimal punctuation also splits: “3.5” counts as two. Unicode text is normalized to NFC. Every passage count and pasted-text estimate uses this same rule.

Timing and interruptions

Text appears when measurement begins. A monotonic clock measures elapsed time; interval callbacks only update the display. Leaving the tab, navigating away, restoring an active reading page, or a clock/sampling gap invalidates the attempt. A plain focus change without hiding the page does not.

Our engineering checks reject readings shorter than 10 seconds, longer than 30 minutes, or outside 1–1,000 words/min. A sampling gap over five seconds or a clock discrepancy over two seconds also invalidates timing. These are product-quality guards, not scientific boundaries for human ability.

The one-minute option

At 60 seconds, reading stops and you select the last fully completed word. Time spent selecting is excluded. An incomplete passage gives a self-reported speed result with comprehension not measured. If you finish the whole passage early, actual time is used and you can choose the optional five-question check.

A personal baseline

Three distinct informational passages, read fully with no recorded prior exposure, can form a baseline. We report the median pace, minimum and maximum, with optional question scores shown separately. A combined score out of 15 appears only if all three checks were submitted. Questions are not required for the pace summary. Repeated, interrupted, anomalous, partial, and one-minute readings are excluded.

Exposure records are limited to this browser. Clearing data, using another device, or blocked storage can remove that knowledge. Similar passage length does not mean equivalent difficulty. This summary is not psychometrically validated.

Where the content comes from

All twelve passages and their questions were created using OpenAI’s AI tools to provide clear, easy-to-follow reading practice. They are AI-generated practice materials reviewed by Codex AI for passage/key consistency. Fiction and explanatory scenarios are invented. No independent human expert review or difficulty calibration is claimed. Six texts are reserved for the timed tool; six different resources support open-book practice and printing. Practice does not consume the timed pool.

Interpret with care

Purpose, text, language familiarity, attention, and prior exposure all matter. This self-check does not diagnose literacy, estimate IQ or reading age, assign percentiles, or certify improvement. Read the research context, or try a natural-paced passage.