Which tasks did Claude Mythos 5 improve quantum circuits for?

Claude Mythos 5 worked on eight quantum circuit tasks covering numerical computation, chemistry, optimization, quantum simulation and number theory. Highlighted results include elliptic-curve point addition, a FeMoco chemistry estimate and a decoder for decoded quantum interferometry. The arXiv paper says seven tasks cut both Toffoli and qubit counts, and the eighth found a new reversible modular-inversion architecture.

Did humans guide Claude during the quantum circuit search?

No. Anthropic says Claude Mythos 5 worked in a closed loop with no human input between rounds. An automated evaluator checked each proposed circuit for correctness and scored it on qubits times gates. Humans wrote the plain-language task specifications and the evaluator. The paper describes this as an evaluator-guided evolutionary framework for quantum circuit design.

Are Claude's quantum circuit designs always smaller than earlier ones?

Not on every measure. One of Claude Mythos 5's elliptic-curve point-addition designs uses 12% more qubits than the smallest previous design but 17 times fewer gates. Its decoded quantum interferometry decoder also fails in about one run in 10,000 when it exceeds reserved space, which Anthropic says is within the task's allowed error rate.

Where can I read the full Claude quantum circuits paper?

The full paper is 'Discovering state-of-the-art quantum circuits with large language models' by Andi Gu, Alek H. Kemeny, Susanne F. Yelin and Mikhail D. Lukin, posted to arXiv as 2610.22367 in the quant-ph category on October 15, 2026. Anthropic's research post summarises the main results, and no code repository is linked.

Anthropic · 2026-10-16 · major

Claude Mythos 5 designs quantum circuits — beats the best published on eight tasks

Anthropic and Harvard researchers ran Claude Mythos 5 in a closed loop with an automated checker to design quantum circuits. Its designs beat the best published circuits on all eight tasks, from chemistry to number theory.

Anthropic research post artwork for designing more efficient quantum circuits with Claude

Claude Mythos 5 searched for smaller quantum circuits and cut gate counts on chemistry, optimization and number-theory tasks.

Quick facts

Model	Claude Mythos 5
Tasks improved	8 of 8 vs best published circuits
Method	LLM proposals + automated evaluator loop
FeMoco Toffoli estimate	~340 million (was ~50 trillion in 2017)
Collaborators	Harvard University, QuEra Computing
Paper	arXiv 2610.22367

What is it?

Anthropic reports that Claude Mythos 5 improved on the best published quantum circuits for eight tasks, using fewer qubits and fewer gates. The work was led by Andi Gu in Anthropic's STEM Fellows Program, with Alek Kemeny and Harvard's Susanne Yelin and Mikhail Lukin, and is written up in the arXiv paper 'Discovering state-of-the-art quantum circuits with large language models'.

How does it work?

The team gave Claude plain-language specifications for each circuit. An automated evaluator checked correctness and scored designs on qubits times gates, and Claude kept proposing new versions with no human input between rounds, an evolutionary search. One elliptic-curve point-addition design came from a single 53-hour run and uses a new reversible way to do divsteps-based modular inversion.

Why does it matter?

Smaller circuits bring useful quantum computing closer. The FeMoco chemistry estimate fell from about 50 trillion Toffoli gates (2017) to about 340 million, roughly a few hours on a machine with about three million physical qubits. A decoded quantum interferometry decoder needs 12 times fewer gates than Google's earlier design, and one point-addition circuit uses 17 times fewer gates.

Who is it for?

quantum algorithm researchers and AI-for-science teams

Frequently asked questions

Which tasks did Claude Mythos 5 improve quantum circuits for?: Claude Mythos 5 worked on eight quantum circuit tasks covering numerical computation, chemistry, optimization, quantum simulation and number theory. Highlighted results include elliptic-curve point addition, a FeMoco chemistry estimate and a decoder for decoded quantum interferometry. The arXiv paper says seven tasks cut both Toffoli and qubit counts, and the eighth found a new reversible modular-inversion architecture.
Did humans guide Claude during the quantum circuit search?: No. Anthropic says Claude Mythos 5 worked in a closed loop with no human input between rounds. An automated evaluator checked each proposed circuit for correctness and scored it on qubits times gates. Humans wrote the plain-language task specifications and the evaluator. The paper describes this as an evaluator-guided evolutionary framework for quantum circuit design.
Are Claude's quantum circuit designs always smaller than earlier ones?: Not on every measure. One of Claude Mythos 5's elliptic-curve point-addition designs uses 12% more qubits than the smallest previous design but 17 times fewer gates. Its decoded quantum interferometry decoder also fails in about one run in 10,000 when it exceeds reserved space, which Anthropic says is within the task's allowed error rate.
Where can I read the full Claude quantum circuits paper?: The full paper is 'Discovering state-of-the-art quantum circuits with large language models' by Andi Gu, Alek H. Kemeny, Susanne F. Yelin and Mikhail D. Lukin, posted to arXiv as 2610.22367 in the quant-ph category on October 15, 2026. Anthropic's research post summarises the main results, and no code repository is linked.