How GPT-5.6 Sol helps run quantum computing experiments

According to OpenAI, Beatriz Yankelevich, a graduate student in MIT’s Engineering Quantum Systems Group, tested GPT-5.6 Sol harnessed to Codex to explore streamlining quantum experimental workflows. Studying superconducting qubits cooled inside dilution refrigerators, Yankelevich connected Codex to lab software to enable the AI to run measurements, analyze results, and determine next steps. GPT-5.6 Sol completed routine measurement workflows autonomously on an uncalibrated six-qubit chip, identifying transition frequencies, calibrating control and read pulses, and determining how long qubits retained quantum information. When signals were clear, Codex finished standard measurement sequences with minimal researcher intervention. However, GPT-5.6 Sol encountered difficulties when experimental signals were weak or noisy, taking longer to find parameters and occasionally needing guidance from an experienced researcher. EQuS now regularly uses agents for routine measurements, which take days to characterize per chip. The agents allow researchers to run measurements overnight or while working elsewhere, checking in via phone. For novel experiments, Yankelevich assigns Codex agents narrower goals and uses their ability to write, modify, and test new code. Yankelevich states that she has built infrastructure to guide agents through measurement, theory, and chip design, allowing her to deploy multiple agents simultaneously and focus on higher-level work like interpreting results, devising experiments, and planning next steps.