Rug placement test

Test one different room. Placement and furniture detection are separate paid steps. Your first room stays saved. Each step permits one attempt, including a failed result.

Placement uses GPT-6 Astra with up to 4,096 output tokens: up to about US$0.21 for output at current standard rates, plus photo and prompt input charges. Furniture detection has its own paid button after a successful placement. No automatic retries.

Test second-room placement — 1 paid attempt

If the saved placement is unsuitable, you can test the updated instructions once. This sends this photo to OpenAI again and uses one remaining paid attempt. Better placement is not guaranteed.

Test updated instructions — 1 attempt

Compare GPT-6 Astra with your saved result using the same photo and placement instructions. This is one additional paid request beyond the three earlier tests. It uses prepaid API credit and may still place the rug incorrectly. Loading a saved result is free; only the comparison button starts this new AI request.

This comparison allows up to 4,096 output tokens, including reasoning. At current standard rates, that is up to about US$0.21 for output, plus the photo and prompt input charges. There is no automatic retry.

Compare with GPT-6 Astra — 1 paid attempt

The earlier Astra result is still shown. The revised instructions allow estimated rug corners beyond the photo edge. This tests a possible explanation for the decline; better placement is not guaranteed.

This is one additional paid GPT-6 Astra request beyond the four earlier tests. It uses the same photo, low reasoning effort and up to 4,096 output tokens: up to about US$0.21 for output at current standard rates, plus photo and prompt input charges. There is no automatic retry.

Test cropped-rug support — 1 paid attempt

Choose a room photo to begin.