Released: 2026-06-03
Missing Domino:
meta-llama/llama-4-maverick
meta
The image contains 27 dominoes, and upon inspection, it is evident that all possible pairs from 0 to 6 are represented except for the double zero (0, 0).
anthropic/claude-haiku-4.5
anthropic
By systematically checking all 28 possible dominoes in a standard double-six set (pairs of numbers from 0 to 6), I identified 27 dominoes in the image. The domino with 0 dots on one side and 1 dot on the other side (the [0,1] domino) is the only one missing from the complete set.
google/gemini-3-flash-preview
After cataloging all 27 dominoes shown in the image, the set represents a double-six set (28 tiles total). The visible tiles include pairs for all combinations except for the double-six (6, 6). All other doubles [0,0], [1,1], [2,2], [3,3], [4,4], and [5,5] are present.
anthropic/claude-sonnet-4.6
anthropic
By systematically cataloging all visible dominoes in the image and comparing against a complete standard double-six set (28 tiles), I identified all present dominoes. The complete set includes every combination from [0,0] to [6,6]. After mapping out all 27 visible tiles, the missing tile is [0,3] (also written as [3,0]).
openai/gpt-5-image-mini
openai
Every other tile of the double-six set appears in the image; the only double missing is the 6-6 tile.
moonshotai/kimi-k2.5
moonshot
By analyzing the 27 visible dominoes and counting the occurrences of each number (0-6), I found that the number 0 appears 7 times and the number 5 appears 7 times, while all other numbers appear 8 times. In a complete double-six set, each number appears exactly 8 times across the 28 dominoes. The missing domino must be the one containing both numbers that appear only 7 times. Visible dominoes with 0 include: [0,0], [0,1], [0,2], [0,3], [0,4], and [0,6]. Visible dominoes with 5 include: [1,5], [2,5], [3,5], [4,5], [5,5], and [5,6]. The only combination not present is [0,5].
Each model receives the same domino puzzle grid. Models are tasked with identifying the missing domino using only visual clues from the grid. Their answers are evaluated without feedback or retries.