How we test

This site exists because air fryer cook times by exact model are scattered across hundreds of PDF manuals, never compared with each other, and often copied from site to site with no verification. Here's exactly how we build every piece of data we show.

1. Official sources only

Every measurement comes from the manufacturer's official instruction manual or official product page — never from a third-party blog, never from a forum. We never republish the PDF itself: we cite the exact page of the cooking chart the data comes from.

2. Nothing is invented

If a manual gives no time for a food on a given appliance, we don't show one — we say so. If it gives a weight range instead of a single value, the range stays as-is; we don't average it into a figure that appears in no manual. Where a spec sheet states the wattage as a range — Tower's T17088 manual says "2200-2600 W" — the headline number is the manufacturer's own upper bound, the one its product page and every retailer listing use, and the full range is quoted verbatim in the source note on that appliance's page.

3. Provenance is always visible

Every measurement carries a label: manufacturer data (from the manual or product page, not independently verified by us), tested by us (measured directly), or reader contribution (submitted by a reader, under moderation before publication). That label sits next to the measurement itself, never hidden in a tooltip.

4. Food safety comes exclusively from the USDA

The internal-temperature thresholds we show (165°F for poultry, 160°F for ground meat, 145°F + 3 minutes' rest for whole cuts of beef/pork/lamb and for fish) come only from USDA guidance — never from a manufacturer manual — and they never change from one appliance to another. A manufacturer may recommend a different cook temperature; the safety threshold we show always stays the same.

5. Disagreements are shown, not smoothed over

Two manufacturers can give very different times for the same food at the same temperature — see how far two official charts diverge on the same chicken breast. We never average two appliances together: every measurement keeps its own source. Where our own test and a manufacturer's figure for the same appliance and food differ by more than 2 minutes, the food page flags both.

6. Bad data fails the build

Every record passes a strict schema before it can publish. A missing required field, or a safety threshold that isn't one of the USDA's own values, stops the build outright rather than shipping an incomplete or unsafe page.