nabu.science
MethodologyAboutContact

 

  • Snapshot
  • Key Claims
  • Strengths & Limitations
  • Commentary
  • From the Authors
  • Author Note / Field Feedback
  • Dimension breakdown
  • Impact Profile
Free · About 5 minutes
Report #NBU-3143049B·Framework v5.13·Pipeline v0.4·Evaluated 22 Nov 2026·OECD: 1.3 Physical sciences·Subfield: Artificial Intelligence·Basis: Full text·Type: Laboratory experiment

High-fidelity entangling gates and nonlocal circuits with neutral atoms

2026·Science·10.1126/science.aei1312 ↗

Primary claim

Using a neutral-atom quantum processor, we realized entangling controlled- Z gates with a high-Rabi-frequency smooth-amplitude pulse, employing state-selective readout and qubit reuse for fast calibration, and achieved a fidelity of 99.854(4)%, which improved to 99.941(3)% upon loss postselection

Supported · orthogonal benchmarks agree, but incompletely mapped leakage can bias the quoted fidelity beyond its statistical uncertainty
In brief

Neutral-atom CZ gates reach 99.854% raw fidelity with extensive cross-checks, but circuit-level validation remains indirect.

ReliabilityCaveats3 of 4 checks completeRead the entry ↓
Likely to hold — the claim's weakest requirement is that the effect is as large as stated. Based on how the study was designed and conducted, not on statistics extracted from the paper. Experimental. Every point below is traceable to a quoted passage, but this forecast has not yet been validated against known replication outcomes.
Abstract

The generation and manipulation of entanglement with low error are essential in quantum information systems. In practice, two-qubit entangling gates constitute a dominant error source, limiting circuit depths and performance in fault-tolerant architectures. Using a neutral-atom quantum processor, we realized entangling controlled- Z gates with a high-Rabi-frequency smooth-amplitude pulse, employing state-selective readout and qubit reuse for fast calibration, and achieved a fidelity of 99.854(4)%, which improved to 99.941(3)% upon loss postselection, with stable performance for 10 hours. We then used these low-error gates in quantum circuits with coherent atom rearrangement. Performance was benchmarked by creating and disentangling cluster states, and subsequently, we studied nonlocally entangled states with scrambling circuits featuring longer-range connectivity. Our approach provides a route toward deep-circuit, efficient fault-tolerant quantum computation.

Key Claims

Primary finding · Supported

Using a neutral-atom quantum processor, we realized entangling controlled- Z gates with a high-Rabi-frequency smooth-amplitude pulse, employing state-selective readout and qubit reuse for fast calibration, and achieved a fidelity of 99.854(4)%, which improved to 99.941(3)% upon loss postselection

Likely to hold — the claim's weakest requirement is that the effect is as large as stated. Based on how the study was designed and conducted, not on statistics extracted from the paper. Experimental. Every point below is traceable to a quoted passage, but this forecast has not yet been validated against known replication outcomes.

orthogonal benchmarks agree, but incompletely mapped leakage can bias the quoted fidelity beyond its statistical uncertainty

Abstract
Quote verified
Secondary · Well-supported

we observe stable performance at ∼99.85% fidelity for over 10 hours without any recalibration of the gate

repeated benchmarking directly tracks the stated operating interval without gate recalibration

Benchmarking low-error CZ gates; Fig. 2D
Quote verified
Secondary · Supported

[In a characterization circuit repeatedly making and disentangling 20-atom cluster states with atom motion] We find a raw fidelity of 99.843(6)% and a loss post-selected value of 99.956(5)%, similar to the fidelity measured in gate benchmarking sequences.

the return-probability estimator is simulation-checked but is not a rigorous gate benchmark

Section 'High-fidelity circuits with atom motion', p. 4, Fig. 4C
Quote verified
Secondary · Supported

in Fig. 5F we compare XEB for the 20-qubit circuit sampling distribution to numerical simulations, and find good agreement for the same error model across three different circuit depths

finite-depth output comparisons support model consistency rather than an independent gate-fidelity measurement

Section 'Chaotic nonlocal circuits', p. 5, Fig. 5E–F
Quote verified
Secondary · Tentative

we design and implement nonlocal circuits featuring longer-range atom transport, which exhibit super-ballistic scrambling of quantum information.

finite-size output sampling does not directly establish the theoretically analysed entropy-growth scaling

Introduction
Quote verified

Derived from the full evaluation — not a separate score.

Strengths · Limitations

Strengths

#methodological rigour

Complementary benchmarks cross-check the central gate result

Echo randomized benchmarking and symmetric stabilizer benchmarking give consistent fidelity estimates. Readout comparisons and interleaved measurements additionally test readout asymmetry, phase errors, and drift.

↳ Fig. 2C; Figs. S2–S3

#contribution

Higher fidelity persists over extended operation

The reported 99.854(4)% raw and 99.941(3)% loss-postselected fidelities exceed the cited neutral-atom benchmarks. Repeated measurements track approximately 99.85% performance for over 10 hours without gate recalibration.

↳ Introduction; Fig. 2B–E

#positioning

Measured error mechanisms inform improvement paths

The physical error model uses independently measured coherence parameters and reproduces responses to deliberate laser perturbations. Its breakdown connects current limitations to specific changes in coherence, unwanted Rydberg coupling, and laser power.

↳ Fig. 3; Tables S1–S2; Figs. S5–S6

Limitations

#methodological rigour

Leakage limits interpretation of quoted precision

Supplement §3.1 estimates that leakage could overstate fidelity by approximately 0.008 percentage points, exceeding the headline raw estimate's 0.004-percentage-point statistical uncertainty. The F=2 leakage manifold is not directly mapped in full.

↳ Supplement §3.1; Fig. S4

#methodological rigour

Circuit evidence stops short of direct validation

The cluster-state fidelity is an approximate, simulation-checked estimator, while nonlocal-circuit performance relies on loss and cross-entropy benchmarking against a noise model. Super-ballistic entropy scaling is established theoretically rather than measured directly in the experiment.

↳ Figs. 4C and 5B–F; Fig. S7; Supplement §§4–5

#reporting

Statistical and model reporting leaves interpretive gaps

The meaning of the parenthetical uncertainties and which nonlocal noise-model parameters were fitted are not specified. Separately, the Supplement §3.1 error components sum to 0.123%, not the stated 0.127%, without changing the model's overall conclusion.

↳ Benchmarking low-error CZ gates; Supplement §§3.1 and 4

Does this match your read? Add your assessment ↓

Nabu Commentary

The central advance is a lower-error neutral-atom CZ gate whose fidelity is checked through complementary benchmarking protocols and readout methods. Long-duration measurements and controlled laser perturbations extend the characterization beyond a single optimized estimate. The possible leakage bias limits interpretation of the quoted numerical precision, while the transported cluster-state and nonlocal circuits provide approximate or model-dependent validation rather than equally direct gate benchmarks. The scrambling analysis establishes theoretical circuit properties, but the experimental output measurements do not directly establish the proposed scaling. A 0.004-percentage-point arithmetic error in the supplementary decomposition warrants a reporting note without overturning the error-model conclusion.

Evaluate another paper

Free · About 5 minutes · Research preview

Author Note / Field Feedback

Nabu’s assessment, alongside the field’s view.

Are you an author of this paper?

Dimension breakdown

Evidence Score

Out of 5
012345
4.1±0.1

Contribution
4.2

The reported raw and loss-postselected CZ fidelities improve on the cited neutral-atom benchmarks, with additional demonstrations of long-duration operation and compatibility with atom motion. The nonlocal-circuit results establish consistency with simulations rather than direct experimental scrambling scaling.

Introduction; Figs. 2B–D, 4C, 5B–F; Supplement §5
Methodological Rigour
4.2

Complementary benchmarking protocols, readout comparisons, interleaved measurements, and controlled perturbations support the central gate result. Incomplete leakage characterization, unspecified uncertainty definitions, and limited noise-model fitting disclosure constrain the precision and independence of the broader validation.

Fig. 2C–F; Figs. S2–S5; Supplement §§3.1 and 4
Reporting
4.0

The progression from gate design through benchmarking to transported circuits is explicit, and the main text distinguishes circuit proxies from rigorous benchmarks. The Introduction's scrambling assertion and the description of passive stability are less precise than the corresponding qualifications elsewhere.

Introduction; High-fidelity circuits with atom motion; Chaotic nonlocal circuits; Supplement §1.1

"not a rigorous benchmarking method"

Positioning
4.0

The paper compares raw and postselected performance with relevant prior benchmarks and traces physical error mechanisms into proposed improvements. Leakage qualifications are more developed in the supplement than in the headline interpretation, and theoretical scrambling scaling is not supported by an equivalent experimental scaling study.

Introduction; Discussion and outlook; Fig. 3; Supplement §§3.1–3.4 and 5

Lower confidence on Positioning — domain match limited.

Reliability

Caveats3 of 4 checks


Results checkNoted

One reported value does not reconcile, but the key claims do not rest on it.

Reporting and transparencyClean

Competing interests are disclosed and data availability is stated. The supplied evidence does not establish a false availability claim; ethics approval is not applicable to these physical experiments.

Conflict of interest: declared · Data availability: declared

Reference IntegrityNot checked

Reference integrity check could not run: no references section detected in extracted text.

0 references in manuscript

Post-Publication RecordClean

No retraction or expression of concern found in Crossref (includes Retraction Watch).

Last checked: 22 Nov 2026

Sources: Crossref (incl. Retraction Watch) ✓

Impact Profile

Impact areaProducts and technology · SDG 9
Type of advanceEmpirical
Route to useSuggested
AdvanceExtends
CitationsToo early
Policy · patentNot yet traced
ReachedNot reachedNot assessable or unverified
ReadinessPiloted
Foundational
Controlled
Piloted
Validated
Implementable
FoundationalImplementable

The gates and circuits were demonstrated working in a single laboratory processor with extensive benchmarking, but with no independent replication or operation in a deployed system.

From the paper
“Our results pave the way for practical fault-tolerant quantum computation with atomic processors.”

The paper names a concrete real-world use of its own findings, namely enabling fault-tolerant quantum computation and high-rate error-correcting codes, but does not name who would act on them.

Pathway

Neutral-atom quantum processor developers would adopt these high-fidelity gate and calibration techniques to reach below-threshold fault-tolerant quantum computation.

Barrier

The gates are demonstrated on at most tens of qubits in one laboratory, and scaling to thousands of qubits with stable, fully error-corrected logical operation remains unshown.

Evaluated 22 Nov 2026 · GBD 2023 · glm_v1_3_3 · impact_profile_v1_3_3

Report an issue →

AI-generated, human-governed. Something look off? Contact us to request a review.

Keep reading

Related evaluations

Other papers Nabu has evaluated blind, starting with the same field.

  1. Retrofitting language models to operate over bytes

    2026 · Computer and information sciences

    Primary claim

    Our byteified models outperformed, on average, all earlier publicly available byte-level LLMs of comparable size.

    Supported
    Evidence Score
    3.9±0.2
    Reliability
    No concerns
  2. FraudBench: A Legal Evaluation of AI Deception

    Wei, Adan, Llerena, et al. · 2026 · Computer and information sciences

    Primary claim

    LLM fraud rates increased with autonomy, with overall rates of 3.0% when advising, 14.7% when drafting, and 21.4% when acting directly, without applying pressure

    Tentative
    Evidence Score
    3.4±0.1
    Reliability
    No concerns
  3. Recursa: Toward Recursively Self-Improving Foundation Model Systems

    Zhu, Lin, Shi, et al. · 2026 · Computer and information sciences

    Primary claim

    On Harvey LAB's Firm Knowledge benchmark, Recursa raises the all-pass rate on held-out test tasks from 30.0% to 55.0% with DeepSeek-V4.1-Flash and from 35.0% to 46.0% with Qwen3.8-Flash.

    Tentative
    Evidence Score
    3.0±0.2
    Reliability
    No concerns

How the rubric works →