Finding
The prompt, v1, frozen
Using LaTeX and TikZ, draw a Venn diagram of three overlapping circles labelled "Has espresso", "Has milk" and "Served cold". Write each of these drinks in the region where it belongs: short black, flat white, iced latte, iced long black, cold brew, babyccino, milkshake, pour-over.
| model | runs | Compiles | Seven regions, three names | Every drink in its region | No label on an edge |
|---|---|---|---|---|---|
| Opus 5.5 | 3 | 3 of 3 | 3 of 3 | 3 of 3 | 3 of 3 |
| Qwen3.8 Flash | 3 | 3 of 3 | 3 of 3 | 3 of 3 | 3 of 3 |
| Gemini 3.8 Flash | 2 | 2 of 2 | 2 of 2 | 2 of 2 | 2 of 2 |
| GPT-6 Sol | 13 | 13 of 13 | 13 of 13 | 13 of 13 | 11 of 13 |
| Claude Sonnet 5.5 | 3 | 3 of 3 | 2 of 3 | 2 of 3 | 2 of 3 |
| DeepSeek 4.1 Flash | 3 | 3 of 3 | 3 of 3 | 3 of 3 | 1 of 3 |
| GLM 5.3 FlashX | 3 | 3 of 3 | 3 of 3 | 3 of 3 | 1 of 3 |
| Grok 4.7 | 3 | 3 of 3 | 3 of 3 | 1 of 3 | 1 of 3 |
| Command A+ | 3 | 3 of 3 | 3 of 3 | 1 of 3 | 0 of 3 |
| Ember 1 | 3 | 3 of 3 | 3 of 3 | 3 of 3 | 0 of 3 |
| GPT-6 Luna | 3 | 3 of 3 | 3 of 3 | 3 of 3 | 0 of 3 |
| Kimi K3 | 3 | 3 of 3 | 3 of 3 | 3 of 3 | 0 of 3 |
| MiMo v2.6 Pro | 2 | 2 of 2 | 2 of 2 | 2 of 2 | 0 of 2 |
| Perceptron Mk1.5 | 3 | 3 of 3 | 3 of 3 | 0 of 3 | 0 of 3 |
Each cell is runs that passed that check, of runs where it was decided. Every run, with pictures.
The edge check was loosened on 2026-09-26 to test the letters themselves, because it had failed labels that sat clear by eye. Initially it compared the size of the text box compared to the circle position. But because of margins it rejected some working ones.
Different strategies to keep the labels clear of the circle line between models with four of the nine runs checking the layout with a calculation, and five of the 22 runs that just looked at the render with vision.
This is the first of four posts on whether measuring helps with achieving this task, in the next post we check this in more details.
The 24 new runs cost $2.07.