We measure ourselves the same way, and publish it.
What we will publish about ourselves, and when
PicksYou is a business on the Growth plan and the weekly scheduler has it. Nothing has been sampled yet, so there is no count on this page and no drawing of one. This is what will be measured.
The questions, as asked
Word for word, in the slot order the plan runs them. No question names PicksYou: a question that carries the name biases the answer and the receipt is then worth nothing.
A question is only ever added at the end of the list, so an earlier one keeps meaning what it meant.
Engines and models
The exact string each vendor's API answered with, and the day it was pinned. A change is a new row here, never an edit.
- ChatGPT gpt-5.4-nano, pinned 2 Sep 2026 times a question: 3. not asked yet
- Perplexity sonar, pinned 27 Aug 2026 times a question: 3. not asked yet
- Google AI Overviews via dataforseo, pinned 27 Aug 2026 times a question: 3. not asked yet
- Grok grok-4.20-0309-non-reasoning, pinned 28 Aug 2026 times a question: 1. not asked yet
- Claude claude-haiku-4-5, pinned 1 Sep 2026 times a question: 1. not asked yet
Official APIs only. No consumer app is read.
PicksYou, measured by PicksYou. The same scheduler, the same matcher and the same interval as every customer's workspace; the figures on this page are read from the stored answers when it is opened. The run starts every Monday at 06:00 UTC; the next one is 7 September. Method: How is this measured?