Skip to content
RESEARCH · QUANTUM SOFTWARE

Pixel-art cats test a quantum computer’s input problem

Moth’s Tessa method spreads data across an IBM chip’s existing connections. Reading it back shows both the promise and the cost.

Published Source report Sources checked
Reading tools for text size, copying, printing and your reading list
My reading list
Reflective metal enclosure labeled IBM Quantum System Two in a research laboratory.
Illustrative photoU.S. Department of State, Foreign Press Centers

Before a quantum computer can analyze an ordinary image, someone must turn its numbers into a quantum state. Moth Quantum presented a new approach on November 16 using a deliberately recognizable test: pixel-art cats that went into IBM processors and came back out.[1]

The method, called Tessa, works with the connections a chip already has. Some carry data; neighboring qubits help identify which values are being addressed. Sharing those neighbors lets the pattern spread across the device without extra operations that move information between distant qubits.[1]

Daniel Bultrini’s November 13 preprint reports encoding and recovering estimates of 1,568 numerical values on IBM’s 120-qubit Phoenix processor. The results are reported in a preprint, not a peer-reviewed paper.[2]

One demonstration used a 28-by-28 image, or 784 pixels. Encoding brightness and hue, two values per pixel, doubled the numerical payload. That larger version was less accurate: the reported correlation between recovered brightness and input fell from 0.96 to 0.69 on Phoenix. Correlation describes how closely the pattern follows the original, not the percentage of correct pixels.[2]

Reading the data also required repeated runs: 8,192 measurements for each measurement basis. The count is not directly comparable to ordinary computer memory. The reconstruction is an estimate, and sampling and hardware noise affect its precision.[2]

The useful question is what comes after loading. Moth says the intended purpose is to process data further on the chip, rather than merely retrieve it. For researchers exploring quantum machine learning, Tessa offers another input method to test. Whether that helps a useful application depends on the whole calculation, including preparation and readout, not on the number of values alone.[1][2]

Keep your reading trail

Evidence & context

2 linked sources

November 13 preprint and November 16 company account; hardware encoding/readout experiment, not demonstrated application advantage.

How to read the evidence and dates

An announcement records what an organization stated. An experiment reports an observation. A simulation or resource estimate depends on its assumptions. QubitWire’s analysis connects these sources; the labels are not quality scores.

Published is the article’s first QubitWire publication. Updated marks a substantive content change. Sources checked records a review of the underlying evidence. An older paper can support a new explanation, and another press release about the same work is not independent replication.

Sources

  1. Moth: Tessa public account, November 16, 2026

    Researcher/company account of the method and intended use; not independent validation.

  2. Daniel Bultrini: Tessa preprint, November 13, 2026

    Version1; Sections2–3 and Table1 contain the encoding, repeated-measurement cost and hardware reconstruction results.

READ NEXT

Continue exploring

All Research stories
  1. Quantum computing / The input problem

    A 20-qubit experiment targets quantum computing’s input problem.

    DLR researchers shortened quantum input-preparation circuits in benchmarks and tested an approach on a real processor. Accuracy and the cost of the whole calculation still decide the payoff.

    Published · Source report · Sources checked · About 3 min read