Quantum computing with realistically noisy devices
Open accessThe DOI matches its Crossref record; 60 of the 60 references it cites were found in a registry or an index; 16 of the 55 sources its claims cite were checked for support.
Abstract
In theory, quantum computers offer a means of solving problems that would be intractable on conventional computers. Assuming that a quantum computer could be constructed, it would in practice be required to function with noisy devices called ‘gates’. These gates cause decoherence of the fragile quantum states that are central to the computer's operation. The goal of so-called ‘fault-tolerant quantum computing’ is therefore to compute accurately even when the error probability per gate (EPG) is high. Here we report a simple architecture for fault-tolerant quantum computing, providing evidence that accurate quantum computing is possible for EPGs as high as three per cent. Such EPGs have been experimentally demonstrated, but to avoid excessive resource overheads required by the necessary architecture, lower EPGs are needed. Assuming the availability of quantum resources comparable to the digital resources available in today's computers, we show that non-trivial quantum computations at EPGs of as high as one per cent could be implemented.
Overview
CoChat Summary
TL;DR
Knill presents the C4/C6 fault-tolerant architecture and evidence that scalable quantum computing may be possible at gate error probabilities above 3% per gate, though resource costs are excessive at such high error rates. At lower error probabilities around 1%, non-trivial computations could be implemented with resources comparable in scale to classical computing.
Key findings
- Postselected C4/C6 computing: heuristic analysis gave thresholds above γ = 0.03; direct simulation up to two levels suggested a threshold above γ = 0.06 by extrapolation, and at γ = 0.03 logical preparation and measurement errors were consistent with being below threshold.
- Error-correcting C4/C6 architecture: threshold above 0.01. At γ = 0.01, logical CNOT detected errors were 2.4 ± 0.0 × 10⁻² at level 3 and 2.4 ± 1.0 × 10⁻³ at level 4; conditional errors were 6.4 ± 0.6 × 10⁻⁴ at level 3 and 0.0 ± 4.4 × 10⁻⁴ at level 4. State-injection conditional logical error was 8.6 ± 0.6 × 10⁻³ at level 3 and 1.1 ± 0.1 × 10⁻² at level 4.
- Resource example: at γ = 0.01, a computation with 100 logical qubits and up to 1000 |π/8⟩-preparations requires 1.23 × 10¹⁴ physical CNOTs, with conditional probability of incorrect output ≈ 0.02.
- Comparison at γ = 10⁻⁴: Steane’s architecture achieves effective logical error ≈ 7 × 10⁻¹² using ≈ 420 physical CNOTs per qubit per gate; C4/C6 achieves detected errors of 5.5 × 10⁻⁹ at level 3 or 6 × 10⁻¹⁴ at level 4 using ≈ 2100 or 2.6 × 10⁴ physical CNOTs per qubit, respectively.
Methods
- Architecture: concatenated error-detecting stabilizer codes C4 (four-qubit code) and C6 (three qubit-pair code), with error-correcting teleportation, transversal Clifford gates, logical Bell state preparation, postselected computing, and state injection/purification for universality.
- Error model: independent Pauli errors with a depolarizing model per gate, setting e_c = γ, e_m = e_p = 4γ/15, and e_h = 4γ/5; additional assumptions include no classical computation error or speed constraint and arbitrary two-qubit gates without delay or additional error.
- Analysis: computer-assisted heuristic threshold analysis, direct stabilizer/Clifford simulation in Octave up to multiple encoding levels, and resource estimates using naive models with Fibonacci-sequence error scaling.
Limitations
- Stated: the error model is idealized and does not match physical qubits and gates, notably coherent rotations, correlations between nearby errors, and leakage; memory error and measurement time are not specified; gates other than measurements are assumed to take the same time; classical computation is assumed error-free and unconstrained; arbitrary two-qubit gates are assumed available without delay or additional error, which the paper calls unrealistic.
- Stated: high-EPG resource requirements are excessive, postselection overheads are extreme, simulations do not constitute full statistical proof because of implicit correlations from the random-number generator, resource models are naive and most reliable for levels < 4, and the threshold ≈ 0.028 obtained from recursion is conjectural.
- Evident: the high-threshold claims rely on heuristic analysis, simulation, and extrapolation rather than mathematical proof, and the resource estimates at high γ depend on unoptimized |π/8⟩-preparation and may be impractical.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
10 Valid · 22 Not valid · 20 Not checked · 203 Own results and interpretations
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“However, in the few cases where experiments with two quantum bits (qubits) have been performed, the EPGs currently achieved are much higher, 3 × 10 -2 or more in ion traps 11, 12 and liquid-state NMR. 13, 14”
What the sources say
- Not checked yetcites [11] [11] Leibfried, D., DeMarco, B., Meyer, V., Lucas, D., Barrett, M., Britton, J., Itano, W. M., Jelenkovi´ c, B., Langer, C., Rosenband, T., and Wineland, D. J. Experimental demonstration of a robust, high-fidelity geometric two ion-qubit phase gate. Nature 422 , 412-415 (2003). - source
- Not checked yetcites [12] [12] Roos, C. F., Lancaster, G. P. T., Riebe, M., H¨ affner, H. , H¨ ansel, W., Gulde, S., Becher, C., Eschner, J., Schmidt-Kaler, F., and Blatt, R. Bell states of atoms with ultralong lifetimes and their tomographic state analysis. Phys. Rev. Lett. , 220402/1-4 (2004). - source
- Not checked yetcites [13] [13] Knill, E., Laflamme, R., Martinez, R., and Tseng, C.-H. An algorithmic benchmark for quantum information processing. Nature 404 , 368-370 (2000). - source
- Supportedcites [14] [14] Childs, A. M., Chuang, I. L., and Leung, D. W. Realization of quantum process tomography in NMR. Phys. Rev. A 64 , 012314/1-7 (2001). - source
The cited paper is a two-qubit liquid-state NMR demonstration (13CHCl3 proton-carbon, Section III Experiment, p. 4) measuring a controlled-NOT gate. Its Results (Gate fidelities, pp. 4-5) give minimum fidelity F_cnot = 0.80 ± 0.04 and average fidelity F_avg_cnot = 0.955, i.e. per-gate error probabilities of ~2×10^-1 and ~4.5×10^-2, both at or above 3×10^-2; the Conclusions (p. 7) state explicitly that these measurements 'suggest a per-gate error rate of the order of 10^-1'. The only better figure mentioned (10^-2) is attributed to another work as a suggestion, not an achieved result, so it does not contradict the claim. The ion-trap half of the sentence is carried by other citations (refs 11,12) and is not judged here. (LLM, conf=0.85)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“This assumption is unrealistic, but the effect on the threshold is due primarily to relatively short-range CNOTs acting within the ancillas needed for maintaining one or two blocks.”
What the sources say
- Partly supportedcites [9] [9] Steane, A. M. Overhead and noise threshold of fault-tolerant quantum error correction. Phys. Rev. A 68 , 042322/1-19 (2003). - sourceExperimental — accuracy not yet measured
In “Timing and non-nearest-neighbour coupling” (p. 7), Steane says the threshold estimates ignore the cost of qubit separation and that most physical gates in the QEC network can connect fairly close qubits. The correction protocol (p. 5) includes CNOTs within ancilla-preparation networks. Our copy does not establish that the threshold effect is due primarily to those CNOTs, specifically within ancillas maintaining one or two blocks. (LLM, conf=0.81)
- Not checked yetcites [20] Steane, A. M. Quantum computer architecture for fast entropy extraction. Quant. Inf. Comput. 4, 297–306 (2002)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
- Partly supported
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“For the C 4 /C 6 architecture, memory is an issue only when waiting for measurement outcomes that determine whether prepared states are good, or that are needed after teleportation, particularly when implementing non-Clifford gates. 18,19”
What the sources say
- Could not verifycites [18] [19] Zhou, X., Leung, D. W., and Chuang, I. L. Methodology for quantum logic gate construction. Phys. Rev. A 62 , 052316/1-12 (2000). - sourceExperimental — accuracy not yet measured
Our copy of Zhou, Leung & Chuang (2000) never addresses quantum memory as an architectural bottleneck: searches for memory/storage of measurement outcomes, feed-forward delay, and a 'C4/C6 architecture' returned no passage making such a statement. The paper's C_k notation is the Clifford hierarchy (Sec. III A, p. 3-4) and its resource discussion is confined to gate counts, measurement type and number of ancilla qubits (Conclusion, p. 11: one-bit teleportation 'only requires projective measurement of Z and as many ancilla qubits as the state to be transformed'; two-bit requires Bell measurement and twice as many ancillas) — it says nothing about when memory is or is not an issue. The closest machinery (classically-controlled corrections conditioned on measurement outcomes, footnote [18] p. 17; ancilla states verified by measuring stabilizers, Sec. III B p. 4 and Appendix B pp. 12-15) supports the ingredients of the claim but not the memory assertion, and no other work is credited for that assertion in our copy. (LLM, conf=0.72)
- Could not verifycites [19] Bollinger, J. J., Heinzen, D. J., Itano, W. M., Gilbert, S. L. & Wineland, D. J. A 303 MHz frequency standard based on trapped Be+ ions. IEEE Trans. Instrum. Meas. 40, 126–128 (1991) sourceExperimental — accuracy not yet measured
Our copy of Bollinger et al. (1991) is entirely about a 303-MHz Be+ Penning-trap frequency standard: the Abstract describes the hyperfine transition, 900 μHz linewidths and Doppler-shift control, and the body sections ('150 000 2sMg+ ions were', '5 parts > 0.8194 T', '550 s were', '3 1 3 - m') cover sympathetic cooling, Ramsey detection and pressure shifts. No passage addresses quantum-computing architectures, C4/C6 codes, memory constraints, state-preparation verification, teleportation or non-Clifford gates; searches for those terms ('quantum logic gate computation qubit architecture', 'memory storage of quantum states long coherence time', 'waiting for measurement outcomes prepared states verification') returned only the spectroscopy passages above. Nothing in our copy supports or contradicts the attributed point, so it cannot be audited against this source. (LLM, conf=0.80)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
- Could not verify
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“To complete the G min gate set so that we can implement arbitrary quantum computations, it suffices to add HAD and preparation of the state | π/ 8 〉 = cos( π/ 8) | 0 〉 + sin( π/ 8) | 1 〉 . 26, 27”
What the sources say
- Could not verifycites [26] [21] Bennett, C. H., Brassard, G., Crepau, C., Jozsa, R., Peres, A., and Wooters, W. Teleporting an unknown quantum state via dual classical and Einstein-Podolsky-Rosen channels. Phys. Rev. Lett. 70 , 1895-1899 (1993). - sourceExperimental — accuracy not yet measured
The paper’s Abstract (page not provided) describes teleportation using an EPR pair and classical communication, not how to complete a gate set. Searches for Gmin, HAD or Hadamard, and π/8 state preparation found no passage addressing the claimed sufficiency for arbitrary quantum computation. (LLM, conf=0.75)
- Could not verifycites [27] [33] Bennett, C. H., DiVincenzo, D. P., Smolin, J. A., and Wootters, W. K. Mixed state entanglement and quantum error-correcting codes. Phys. Rev. A 54 , 3824-3851 (1996). - sourceExperimental — accuracy not yet measured
The discussion of local operations and bilateral XOR in the paper (p. 2) concerns entanglement purification, not completion of a gate set. Searches for HAD, the |π/8〉 state, and universal quantum computation found no passage addressing the claimed sufficiency of those additions. (LLM, conf=0.76)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“If the EPG is smaller than a threshold, then scalable quantum computing is possible. 5-8”
What the sources say
- Supportedcites [5] [5] Preskill, J. Reliable quantum computers. Proc. R. Soc. Lond. A 454 , 385-410 (1998). - source
Section 1 (p. 2) states the threshold theorem outright: 'if the error rate is below a critical value, then it is possible to do an arbitrarily long quantum computation with a negligible probability of error.' Section 6 (p. 15) gives the mechanism — concatenation drives the error rate arbitrarily small provided the fundamental gate error rate is small enough — and (pp. 16-17) supplies numerical threshold estimates (~10^-3 crude, ~10^-4 for gate errors, ~10^-5 when storage errors dominate). The paper reports and relies on this result itself, so the attribution holds. (LLM, conf=0.92)
- Supportedcites [6] [6] Kitaev, A. Y. Quantum computations: Algorithms and error correction. Russian Math. Surveys 52 , 1191-1249 (1997). - source
The §0 Introduction states the paper's central fault-tolerance result in exactly this form: Shor's (log L)^-c accuracy requirement is strengthened so that 'fixed accuracy is sufficient' to 'carry out arbitrarily long quantum computations involving operations with fixed accuracy', i.e. a constant per-operation accuracy bound suffices for unlimited-length (scalable) computation via the cascade codes of §10. Definition 10.4 formalises this by taking a basis 'whose elements are realized with fixed accuracy δ' and building a polynomial error correction scheme with accuracy exp(-k), and the paper explicitly notes that for classical transformations 'accuracy' means 'probability of error' (§10 Additional information), so the per-gate error-probability phrasing is the paper's own. The claim's word 'threshold' is not used as such in our copy (the related threshold papers by Knill-Laflamme-Zurek [45] appear only in the reference list), but the substance — a constant per-gate error bound below which scalable quantum computing is possible — is the paper's main theorem. (LLM, conf=0.85)
- Not checked yetcites [7] Aharonov, D. & Ben-Or, M. Fault-tolerant quantum computation with constant error. Preprint at http://www.arXiv.org/quant-ph/9906129 (1999).
- Not checked yetcites [8] [8] Knill, E., Laflamme, R., and Zurek, W. H. Resilient quantum computation. Science 279 , 342-345 (1998). - source
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
147 more claims · 6 with their cited sources checked
Create a free account to see all 255 claims and to verify the 39 references still open.
- It may state two separate claims as one. 67
- Probably not a substantial claim. 47
- The quote only partly states this claim. 4
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“One of the most encouraging results for all approaches to quantum computing was the high threshold of 1% recently reported by Knill [35].”
From Optical Quantum Computing (2007) · cites this paper as [35]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ A Explanation of Error-Correcting Teleportation · p. 20
“Consider an n qubit 'input' block carrying l qubits encoded in the stabilizer code for Q , where the block has been affected by errors.”
§ A Explanation of Error-Correcting Teleportation · p. 20
“Let Q be the l × 2 n binary check matrix with entries defining a stabilizer code on n qubits for encoding k = n -l qubits”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ A Explanation of Error-Correcting Teleportation · p. 20
“The stabilizer consists of the system-labeled Pauli products I, σ x ( 2 ) σ x ( 3 ) , σ y ( 2 ) σ y ( 3 ) and σ z ( 2 ) σ z ( 3 ) .”
§ A Explanation of Error-Correcting Teleportation · p. 20
“The global initial state can be viewed as | ψ 〉 encoded in the stabilizer code generated by IXX and IZZ , whose check matrix has rows b 1 = [001010] and b 2 = [000101] .”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ G Resource Usage · p. 38
“The probability of success is p (1) π/ 8 = 1 -35 ǫ π/ 8 -O ( ǫ 2 π/ 8 ) .”
§ G Resource Usage · p. 38
“Using the exact formulas, for ǫ π/ 8 = 0 . 01 , 0 . 001 , 0 . 0001 we obtain ǫ (1) π/ 8 = 3 . 6 × 10 -5 , 3 . 5 × 10 -8 , 3 . 5 × 10 -11 and p (1) π/ 8 = 0 . 860 , 0 . 985 , 0 . 999 .”
Citations
Cites
Quantum Computation and Quantum Information
Nielsen, M. A., Chuang, I., & Grover, L. K. (2002) · American Journal of Physics · doi:10.1119/1.1463744 · Cited by 21,895
Quantum computation and quantum information
Law, J. (2001) · ACM SIGSOFT Software Engineering Notes · doi:10.1145/505482.505499 · Cited by 18,842
Teleporting an unknown quantum state via dual classical and Einstein-Podolsky-Rosen channels
Bennett, C. H., Brassard, G., Crépeau, C., Jozsa, R., Peres, A., & Wootters, W. K. (1993) · Physical Review Letters · doi:10.1103/physrevlett.70.1895 · Cited by 13,703
Polynomial-time algorithms for prime factorization and discrete logarithms on a quantum computer
Shor, P. W. (1997) · SIAM Journal on Computing · doi:10.1137/s0097539795293172 · Cited by 8,472
Simulating physics with computers
Feynman, R. P. (1982) · International Journal of Theoretical Physics · doi:10.1007/bf02650179 · Cited by 7,691 · cited 8×
“One is encoded into the [[7, 1, 3]] code44 (a code that encodes 1 qubit in7 qubits with minimum distance 3, which implies that it can correct any(3 − 1)/2 = 1 qubit error or detect any (3 − 1) = 2 qubit errors).”
BackgroundMethodInfluentialMixed-state entanglement and quantum error correction
Bennett, C. H., DiVincenzo, D. P., Smolin, J. A., & Wootters, W. K. (1996) · Physical Review A · doi:10.1103/physreva.54.3824 · Cited by 5,272
Raussendorf, R., & Briegel, H. J. (2001) · Physical Review Letters · doi:10.1103/physrevlett.86.5188 · Cited by 4,477
Purification of Noisy Entanglement and Faithful Teleportation via Noisy Channels
Bennett, C. H., Brassard, G., Popescu, S., Schumacher, B., Smolin, J. A., & Wootters, W. K. (1996) · Physical Review Letters · doi:10.1103/physrevlett.76.722 · Cited by 2,961
‘‘Event-ready-detectors’’ Bell experiment via entanglement swapping
Żukowski, M., Zeilinger, A., Horne, M. A., & Ekert, A. K. (1993) · Physical Review Letters · doi:10.1103/physrevlett.71.4287 · Cited by 2,045
Gottesman, D., & Chuang, I. L. (1999) · Nature · doi:10.1038/46503 · Cited by 1,860 · cited 1×
“The principle is as explained in [18] for unitary gates, but generalized to measurements.”
Background
Cited by
Ladd, T. D., Jelezko, F., Laflamme, R., Nakamura, Y., Monroe, C., & O’Brien, J. L. (2010) · Nature · doi:10.1038/nature08812 · Cited by 3,592
Linear optical quantum computing with photonic qubits
Kok, P., Munro, W. J., Nemoto, K., Ralph, T. C., Dowling, J. P., & Milburn, G. J. (2007) · Reviews of Modern Physics · doi:10.1103/revmodphys.79.135 · Cited by 2,904
“General fault-tolerant thresholds for quantum computing have been derived by Steane (2003) and Knill (2005), and here we address LOQC specific error correction and fault-tolerant thresholds.”
MethodQuantum information with Rydberg atoms
Saffman, M., Walker, T. G., & Mølmer, K. (2010) · Reviews of Modern Physics · doi:10.1103/revmodphys.82.2313 · Cited by 2,836
“001, which corresponds to recent theoretical estimates(Aliferis and Preskill, 2009; Knill, 2005) for the fault tolerant threshold in a scalable quantum computer, can be achieved over a large range of interatomic separations.”
InfluentialQuantum error correction for quantum memories
Terhal, B. M. (2015) · Reviews of Modern Physics · doi:10.1103/revmodphys.87.307 · Cited by 1,354
“We take a small stabilizer code C 6 [defined in Knill (2005)] with parity checks X 1 X 4 X 5 X 6 , X 1 X 2 X 3 X 6 , Z 1 Z 4 Z 5 Z 6 , and Z 1 Z 2 Z 3 Z 6 acting on six qubits.”
InfluentialChitambar, E., & Gour, G. (2019) · Reviews of Modern Physics · doi:10.1103/revmodphys.91.025001 · Cited by 1,276
“In fact, this technique is universal in that any non-stabilizer operation can be implemented in such a way, provided the non-stabilizer state consumed belongs to the class of socalled magic states (Bravyi and Kitaev, 2005; Gottesman and Chuang, 1999; Knill, 2005).”
Background
Metadata
- Authors
- Knill, E.
- Affiliations
- National Institute of Standards and Technology (US)
- Corresponding author
- Knill, E.
- ORCID
- Knill, E.
- Venue
- Nature
- Published
- 2005-03
- Volume
- 434
- Issue
- 7029
- Pages
- 39-44
- ISSN
- 0028-0836 (ISSN-L)
- 1476-4687
- OpenAlex
- W2032147770
- Semantic Scholar
- 5d1ac222556297a0a79eb0cc815a05600a168446
- Cited by
- 1,184
- References
- 54
- Citations by year
- 2026: 46 · 2025: 102 · 2024: 78 · 2023: 66 · 2022: 69 · 2021: 81 · 2020: 74 · 2019: 48 · 2018: 40 · 2017: 49 · 2016: 52 · 2015: 39 · 2014: 44 · 2013: 41 · 2012: 36
- FWCI
- 54.93
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- www.springer.com/tdm (tdm)
- Locations
- Nature (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- CERN Document Server (European Organization for Nuclear Research) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Quantum Computing Algorithms and Architecture
- Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum Information and Cryptography
- Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum many-body systems
- Keywords
- fault-tolerant quantum computing, decoherence, quantum computing architecture, resource overhead, quantum error correction
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-07, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-07 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash and gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.