Logical Qubit
A logical qubit is one protected qubit encoded across many error-prone physical qubits. In December 2024, Google's Willow chip showed for the first time that enlarging the code makes the logical qubit better, not worse — the first below-threshold surface code.
What it means
Physical qubits fail roughly once per thousand operations — far too often for useful algorithms.Quantum error correction spreads one qubit's worth of information across many physical qubits, forming a logical qubit whose errors are continuously detected and corrected.The decisive question is whether adding more physical qubits actually helps.Google's Willow chip answered yes for the first time (Acharya et al., Nature 638, 920–926, Dec 2024): each step up in surface-code distance suppressed the logical error rate by a factor Λ = 2.14 ± 0.02, and a distance-7 logical qubit lived 291 µs — 2.4× longer than the best physical qubit on the chip.The encoded whole outlived its best part.The authors' own caveat matters: orders of magnitude remain between present logical error rates (~1.4e-3) and requirements for practical quantum computation (~1e-6 to 1e-10).Counting logical qubits is also subtler than headlines suggest.Quantinuum's Helios (Nov 2025), with 98 physical trapped-ion qubits, reported 94 error-DETECTED logical qubits in a GHZ state and 48 fully error-corrected logical qubits at 2:1 encoding — but the 94 used low-distance 'iceberg' codes relying on postselection (discarding runs where an error was flagged).As critics put it: logical qubit count is not one number — encoding rate, distance, and fault-tolerance level all matter.A third path: the Harvard/MIT/QuEra 448-atom neutral-atom architecture (Bluvstein et al., Nature 649, 39–46, online Nov 2025) integrates the key ingredients of fault tolerance in one system.Everyday analogy
Common misconceptions
- A logical qubit is NOT a better physical qubit — it is information spread across many physical qubits. Comparing a chip's raw physical-qubit count against another's logical-qubit count is comparing different things.
- 'X logical qubits' is not one number: error-detected logical qubits (low-distance codes plus postselection, as in Helios's 94) are a much weaker claim than fully error-corrected, high-distance logical qubits. Encoding rate, distance, and fault-tolerance level all matter.
- Below threshold does not mean solved: Willow's authors themselves note that orders of magnitude remain between present logical error rates (~1.4e-3) and practical requirements (~1e-6 to 1e-10).
Key takeaways
- Many physical qubits plus continuous error correction encode one protected logical qubit.
- Google Willow (Nature 638, Dec 2024) was the first below-threshold surface code: error suppression Λ = 2.14 ± 0.02 per distance step, and a distance-7 logical qubit lifetime of 291 µs — 2.4× the best physical qubit.
- The authors' own caveat: orders of magnitude remain between present logical error rates (~1.4e-3) and requirements for practical quantum computation (~1e-6 to 1e-10).
- Logical-qubit counts are not directly comparable: Quantinuum Helios (Nov 2025) reported 94 error-detected (postselected, low-distance 'iceberg' codes) and 48 fully error-corrected logical qubits at 2:1 encoding from 98 physical qubits — rate, distance, and fault-tolerance level all matter.
- Neutral atoms are a third contender: the Harvard/MIT/QuEra 448-atom architecture (Nature 649, Nov 2025) integrates the ingredients of fault tolerance.
Check your understanding
What did Google's Willow chip demonstrate for the first time in December 2024?
- A.A logical qubit with zero errors
- B.That increasing surface-code distance suppresses logical errors (below threshold), with Λ = 2.14 per distance step
- C.The first quantum computer with 1 million qubits
- D.That physical qubits no longer need error correction
Show the answer
Answer: B. That increasing surface-code distance suppresses logical errors (below threshold), with Λ = 2.14 per distance step
Why: Willow was the first below-threshold surface code: each distance step suppressed the logical error rate by Λ = 2.14 ± 0.02, and the distance-7 logical qubit outlived the best physical qubit by 2.4×. It did not eliminate errors — the authors note orders of magnitude still remain to practical requirements.
Why is 'number of logical qubits' not a single comparable number across machines?
- A.Because logical qubits are imaginary
- B.Because encoding rate, code distance, and fault-tolerance level all differ — error-detected (postselected) logical qubits are a weaker claim than fully error-corrected ones
- C.Because only superconducting chips can have logical qubits
- D.Because logical qubits cannot be measured
Show the answer
Answer: B. Because encoding rate, code distance, and fault-tolerance level all differ — error-detected (postselected) logical qubits are a weaker claim than fully error-corrected ones
Why: Quantinuum Helios reported 94 error-detected logical qubits (low-distance iceberg codes relying on postselection) but 48 fully error-corrected ones at 2:1 encoding — the same machine, very different claims. Rate, distance, and fault-tolerance level all matter.
Builds on
Graded 2026-07-10 (human sign-off): timebound, asOfDate 2026-07-10 — logical-qubit milestones are fast-moving hardware records: Willow below-threshold result (Nature 638, Dec 2024, Λ=2.14±0.02, d-7 lifetime 291 µs), Quantinuum Helios counts (Nov 2025, 94 error-detected / 48 error-corrected), and the 448-atom Harvard/MIT/QuEra architecture (Nature 649) will all be superseded; the ~1.4e-3 vs 1e-6–1e-10 gap statement is dated.
Learn it hands-on
This concept is part of a 46-level curriculum with an interactive simulator and Lumen, a tutor whose answers are verified before you see them. Levels 1–5 are free.