How GPT-5.6 Sol helps run quantum computing experiments

An AI agent can help calibrate a quantum chip when it can operate the laboratory's measurement software and use each result to choose the next experiment. OpenAI's case study describes GPT-5.6 Sol in Codex working with an MIT researcher on an uncalibrated six-qubit chip. The researcher supplied instructions for running and evaluating measurements; the agent selected settings, collected data and used the results to decide what to do next. Clear signals allowed standard measurement sequences to proceed with little intervention. Weak or noisy signals sometimes required expert guidance. The case study describes one laboratory setup and gives no measured time saving, so it shows what routine automation can look like rather than how much supervision it removes.