I checked the stitch count before committing to a scarf

This is a calculation test using an imaginary swatch, not a knitted or measured scarf. I ran it in TiinyOS with Qwen3.6-35B-A3B and reasoning on.

A width mistake becomes expensive after a long scarf is already on the needles. I gave TiinyOS a made-up swatch record: 23 stitches and 32 rows per 10 cm after washing, a 24 cm target width, and a pattern that only works with 6k + 2 stitches. Those two edge stitches are already included in the count.

The unconstrained width calculation lands at 55.2 stitches. Tiiny picked 56, the nearest valid count, and showed its neighbors: 50 stitches would make about 21.74 cm; 56 would make about 24.35 cm; 62 would make about 26.96 cm. It did not add the edge stitches twice. For 150 cm of length, it calculated 480 rows, exactly 60 repeats of the eight-row pattern.

I also asked how much yarn to buy. Tiiny left the ball count open because the swatch record had no yarn-usage measurement. That was the right place to stop. Its advice to remeasure the finished swatch made sense, but it added thresholds such as a 2–5% width shift and a 2% row-gauge drift without support from my input. I would not turn those numbers into pattern rules.

I would try a short 56-stitch sample, wash it and measure the width before starting the 150 cm scarf. This single calculation assumes the stated gauge and does not verify the finished dimensions.