The Pulse

QuEra says AI-designed circuits cut a quantum resource measure by up to 39×

QuEra says a research team used Claude and an automated evaluator to design quantum circuits with lower resource estimates. The largest reported reduction was 39-fold on a combined measure of Toffoli gates and logical qubits, not a measured

AI.info Team · Written with AI · fact-checked · 2026-11-10

AI searches for smaller quantum circuits

QuEra Computing said on Nov. 10 that research with Harvard University and Anthropic used AI to design quantum circuits with lower estimated resource requirements. The largest reported result was a 39-fold reduction in a combined measure of gate and qubit resources, the company said. The work is presented as a way to reduce what future fault-tolerant quantum computers may need to run parts of larger algorithms.

The findings are in the preprint Discovering state-of-the-art quantum circuits with large language models, identified as arXiv:2610.22367. The work examined eight computational building blocks used in larger quantum algorithms; seven designs reduced both gate counts and qubit requirements against their baselines. In the eighth, the search produced another approach that offers choices between qubit use and gate count, according to the release.

An evaluator checked each design

The researchers paired Anthropic’s Claude with an automated evaluator that checked circuit correctness and scored each design’s Toffoli-gate count and qubit count. Claude used that feedback to search for and refine circuit designs, the company said. The measure combines two resource estimates: Toffoli gates, which QuEra says can contribute substantially to fault-tolerant computation overhead, and logical qubits, which occupy hardware capacity.

For a fixed-point Gaussian function, the team reported a 39-fold drop in the combined measure, with Toffoli gates falling from about 24,000 to 1,583 and qubits from 362 to 139. For the Coulomb potential, the reported combined-resource reduction was 9.9-fold, with Toffoli gates decreasing from 6,355 to 1,140 and qubits from 252 to 142. The Reed-Solomon decoder, described as the most expensive step in decoded quantum interferometry, used 12 times fewer Toffoli gates, from 5.1 million to about 440,000, and fewer qubits.

The combined measure multiplies Toffoli-gate count by peak logical-qubit count, so the 39-fold result is not a 39-fold speedup, QuEra cautioned. The figures are resource estimates for individual building blocks, not hardware runs, and their effect on complete applications will depend on how the components are used.

A new circuit trade-off

The researchers also reported that Claude helped discover a reversible approach to modular inversion, an operation used in elliptic-curve calculations. In a complete point-addition benchmark, one version used about 17 times fewer gates than the previous design with the fewest qubits, while using about 12% more qubits. The release says the approach reconstructs much of the information usually stored during intermediate steps, offering another way to balance gate counts against qubit requirements.

Search time and scope

The search that produced the new elliptic-curve circuit approach took 53 hours, guided by researcher-defined requirements and an automated evaluator, QuEra said. The company said the results may help researchers explore circuit designs more quickly, while stressing that the effect on complete applications depends on how the optimized components are used. QuEra is building fault-tolerant quantum computers and says its Libra system is targeted for availability on Amazon Braket in 2028.