Skip to content
← All vendors

Quantum scorecard

IBM

Superconducting

IBM ships Nighthawk, 120 physical qubits, announced Nov 2025. Next on its roadmap: Kookaburra, 2026.

Latest activity on record: Oct 16, 2026

Published figures

Each figure names its device and links to where it was published.

Ships today
120physical qubitsNighthawk · Nov 2025 · curated
Two-qubit gate fidelity
99.7%Heron R2 · vendor
Quantum volume
2,048Heron r3 "Pittsburgh" · Wikipedia
Largest device to date
1,121physical qubitsCondor · Dec 2023
Latest round in the ledger
$1BVentureOct 2026 · Inside Quantum Technology

Qubit count over time

20 announced devices, by announcement date. Records such as logical-qubit demonstrations are listed below, not plotted. The line traces the largest device to date.

1101001,00010,000201620182020202220242026announcement datephysical qubits (log scale)IBM Q 5 Tenerife · Jan 2016 · 5 physical qubitsIBM Q 17 · May 2017 · 17 physical qubitsIBM Q 16 Rüschlikon · May 2017 · 16 physical qubitsIBM Q 20 Tokyo · Nov 2017 · 20 physical qubitsIBM Q System One · Jan 2019 · 20 physical qubitsIBM Vigo · Jul 2019 · 5 physical qubitsIBM Ourense · Jul 2019 · 5 physical qubitsIBM London · Sep 2019 · 5 physical qubitsIBM Burlington · Sep 2019 · 5 physical qubitsIBM Essex · Sep 2019 · 5 physical qubitsIBM Q 53 · Oct 2019 · 53 physical qubitsIBM Armonk · Oct 2019 · 1 physical qubitIBM Hummingbird · Aug 2020 · 65 physical qubitsEagle · Nov 2021 · 127 physical qubitsOsprey · Nov 2022 · 433 physical qubitsIBM Heron · Dec 2023 · 133 physical qubitsCondor · Dec 2023 · 1,121 physical qubitsHeron R1 · Dec 2023 · 133 physical qubitsHeron R2 · Nov 2024 · 156 physical qubitsNighthawk · Nov 2025 · 120 physical qubitsCondor · 1,121Nighthawk · 120

Roadmap

Targets the vendor has published. A target is a promise, not a result.

  1. 2026Kookaburra — modular 4,158-qubit systemnext
  2. 2029Starling — 200 logical qubits, 100M gates
  3. 2033Blue Jay — 2,000 logical qubits, 1B gates

Every milestone on record

20 entries, newest first.

DateDevice
Nov 2025NighthawkCommercial delivery: 218 tunable couplers, circuits up to ~5,000 two-qubit gates.120 physical qubitscuratedSource — Nighthawk
Nov 2024Heron R2Updated Heron with lower crosstalk and larger circuit depth.156 physical qubitscuratedSource — Heron R2
Dec 2023CondorIBM's first >1000-qubit processor.1,121 physical qubitscuratedSource — Condor
Dec 2023Heron R13–5× lower error rate than Eagle; the utility-scale successor.133 physical qubitscuratedSource — Heron R1
Dec 2023IBM Heron133 physical qubitsWikipediaSource — IBM Heron
Nov 2022Osprey433 physical qubitscuratedSource — Osprey
Nov 2021EagleFirst >100-qubit superconducting processor.127 physical qubitscuratedSource — Eagle
Aug 2020IBM Hummingbird65 physical qubitscuratedSource — IBM Hummingbird
Oct 2019IBM Armonk1 physical qubitWikipediaSource — IBM Armonk
Oct 2019IBM Q 5353 physical qubitsWikipediaSource — IBM Q 53
Sep 2019IBM London5 physical qubitsWikipediaSource — IBM London
Sep 2019IBM Burlington5 physical qubitsWikipediaSource — IBM Burlington
Sep 2019IBM Essex5 physical qubitsWikipediaSource — IBM Essex
Jul 2019IBM Vigo5 physical qubitsWikipediaSource — IBM Vigo
Jul 2019IBM Ourense5 physical qubitsWikipediaSource — IBM Ourense
Jan 2019IBM Q System OneFirst integrated commercial quantum system.20 physical qubitscuratedSource — IBM Q System One
Nov 2017IBM Q 20 Tokyo20 physical qubitsWikipediaSource — IBM Q 20 Tokyo
May 2017IBM Q 1717 physical qubitsWikipediaSource — IBM Q 17
May 2017IBM Q 16 Rüschlikon16 physical qubitsWikipediaSource — IBM Q 16 Rüschlikon
Jan 2016IBM Q 5 Tenerife5 physical qubitsWikipediaSource — IBM Q 5 Tenerife

curated: from the code-reviewed milestone list, sourced to the vendor · reviewed: a news-surfaced milestone an editor published · Wikipedia: synced from Wikipedia's list of quantum processors.

System benchmarks on real hardware

metriq-gym runs from the Unitary Foundation's metriq-data dataset, newest run per device. Each benchmark has its own direction; scores do not compare across benchmarks.

  • BSEQ — largest connected qubit set (Metriq)

    higher is better · qubits

    IBM Boston
    156Dec 9, 2025
    IBM Fez
    155Oct 19, 2025
    IBM Pittsburgh
    152Oct 19, 2025
    IBM Marrakesh
    149Oct 19, 2025
    and 4 more devices — see Field data
  • CLOPS — circuit layer operations per second (Metriq)

    higher is better · clops

    IBM Phoenix
    2,551,868Sep 11, 2026
    IBM Fez
    368,905Feb 17, 2026
    IBM Kingston
    362,372Feb 17, 2026
    IBM Boston
    360,574Feb 9, 2026
    and 4 more devices — see Field data
  • EPLG — error per layered gate (Metriq)

    lower is better · eplg

    IBM Kingston
    5.49e-3Aug 20, 2026
    IBM Boston
    7.36e-3Feb 18, 2026
    IBM Pittsburgh
    0.0203Feb 19, 2026
    IBM Phoenix
    0.026Sep 11, 2026
    and 4 more devices — see Field data
  • Linear Ramp QAOA at 20 qubits (Metriq)

    higher is better · approx-ratio

    IBM Boston
    0.6641Nov 26, 2025
    IBM Phoenix
    0.625Sep 11, 2026
    IBM Pittsburgh
    0.6222Feb 17, 2026
    IBM Miami
    0.6058Apr 16, 2026
    and 4 more devices — see Field data
  • Mirror circuits polarization, 16 qubits × 32 layers (Metriq)

    higher is better · polarization

    IBM Boston
    0.4952Dec 16, 2025
    IBM Marrakesh
    0.4846Dec 16, 2025
    IBM Pittsburgh
    0.4613Dec 16, 2025
    IBM Phoenix
    0.4056Sep 11, 2026
    and 4 more devices — see Field data
  • QML Kernel accuracy at 20 qubits (Metriq)

    higher is better · accuracy

    IBM Boston
    0.742Dec 8, 2025
    IBM Kingston
    0.6Nov 11, 2025
    IBM Phoenix
    0.584Sep 11, 2026
    IBM Torino
    0.488Nov 11, 2025
    and 5 more devices — see Field data
  • WIT witness expectation at 7 qubits (Metriq)

    higher is better · expectation

    IBM Phoenix
    0.9169Sep 11, 2026
    IBM Boston
    0.8972Dec 16, 2025
    IBM Pittsburgh
    0.8882Feb 10, 2026
    IBM Miami
    0.8835Apr 16, 2026
    and 5 more devices — see Field data

Funding rounds in the ledger

Extracted from news coverage and de-duplicated; amounts as published.

DateRoundAmountSource
Oct 15, 2026VentureIBM$1BInside Quantum Technology ↗

Recent headlines

Not on record for IBM yet: a live fleet API.

Quantum · Timeline

IBM, in order

Everything the site tracks for IBM — hardware, records, advantage claims it took part in, funding and its own roadmap — newest first.

Newest on the timeline

Promised next · 3

Roadmap targets whose year has not arrived yet, at the vendor's own date. They move into the timeline when the year passes, as shipped or as past their date.

  1. RoadmapSuperconductingPromised for 2026

    IBM · Kookaburra — modular 4,158-qubit system

  2. RoadmapSuperconductingPromised for 2029

    IBM · Starling — 200 logical qubits, 100M gates

  3. RoadmapSuperconductingPromised for 2033

    IBM · Blue Jay — 2,000 logical qubits, 1B gates

20267 events

  1. Advantage claim · Classically verifiable

    ibm_boston_dcs_nq70_depth70_checks27

    0.3%score on this instance (higher is better)

    Instance doped_random_graph_sampling_nq70_depth70_checks27 · IBM, UChicago

  2. Advantage claim · Observable estimation

    Simulating SU(2) Hadron dynamics on a 60 site lattice using 120 qubits

    60.0054estimated observable, against this instance's reference

    Instance su2_hadron_dynamics_lsh_x_100_meson · Birla Institute of Technology and Science - Pilani, IBM

  3. Advantage claim · Classically verifiable

    ibm_boston_rgs_nq70_depth70_checks27_basis_fstate

    0.4%score on this instance (higher is better)

    Instance random_graph_sampling_nq70_depth70_checks27_basis_fstate · IBM, University of Chicago

  4. Advantage claim · Observable estimation

    Simulating SU(2) Hadron dynamics on a 60 site lattice using 120 qubits

    59.994estimated observable, against this instance's reference

    Instance su2_hadron_dynamics_lsh_x_100_meson · Birla Institute of Technology and Science - Pilani, IBM

  5. Advantage claim · Classically verifiable

    mpo_unswapping_56_1917

    100%score on this instance (higher is better)

    Instance peaked_circuit_P9_Hqap_56x1917 · IBM

  6. Advantage claim · Classically verifiable

    tno_cancellation_49x5072

    100%score on this instance (higher is better)

    Instance peaked_circuit_heavy_hex_49x5072 · IBM

202515 events

  1. Advantage claim · Classically verifiable

    Quantum runtime for peaked circuit 49q x 5072 CZ's on heavy-hex

    100%score on this instance (higher is better)

    Instance peaked_circuit_heavy_hex_49x5072 · IBM

  2. Advantage claim · Observable estimation

    49x1296, ibm_pittsburgh

    0.649estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_49x1296 · IBM, Algorithmiq, Flatiron Institute

  3. Advantage claim · Observable estimation

    49x1296, ibm_boston (layout 2)

    0.657estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_49x1296 · IBM, Algorithmiq, Flatiron Institute

  4. Advantage claim · Observable estimation

    Single_path_MC_70x1872

    0.614estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_70x1872 · University of Chicago, IBM, Algorithmiq, EPFL, Flatiron Institute

  5. Advantage claim · Observable estimation

    Single_path_MC_49x1296

    0.619estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_49x1296 · University of Chicago, IBM, Algorithmiq, EPFL, Flatiron Institute

  6. Advantage claim · Observable estimation

    Single_path_MC_49x648

    0.808estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_49x648 · University of Chicago, IBM, Algorithmiq, EPFL, Flatiron Institute

  7. Advantage claim · Classically verifiable

    TNO + Iterative Cancellation for peaked_circuit_heavy_hex_49x4020

    100%score on this instance (higher is better)

    Instance peaked_circuit_heavy_hex_49x4020 · IBM

  8. Advantage claim · Classically verifiable

    Quantum Runtime for peaked_circuit_heavy_hex_49x4020

    100%score on this instance (higher is better)

    Instance peaked_circuit_heavy_hex_49x4020 · IBM

  9. Advantage claim · Observable estimation

    49x648, ibm_boston

    0.824estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_49x648 · IBM, Algorithmiq, Flatiron Institute

  10. Advantage claim · Observable estimation

    49x1296, ibm_boston

    0.662estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_49x1296 · IBM, Algorithmiq, Flatiron Institute

  11. Advantage claim · Observable estimation

    70x1872, ibm_boston

    0.673estimated observable, against this instance's reference

    Instance operator_loschmidt_echo_70x1872 · IBM, Algorithmiq, Flatiron Institute

  12. HardwareSuperconducting

    IBM Nighthawk

    120physical qubits

    Commercial delivery: 218 tunable couplers, circuits up to ~5,000 two-qubit gates.

  13. Advantage claim · Variational

    SKQD

    -62.2567energy in Hartree (lower is better)

    Instance anderson_impurity_model_4i_28b_32e · IBM, ORNL, University of Maryland

  14. Quantum volumeSuperconducting

    IBM Heron r3 "Pittsburgh"

    2,048quantum volume

  15. Quantum volumeSuperconducting

    IBM Heron r3 "Pittsburgh"

    1,024quantum volume

20241 event

  1. HardwareSuperconducting

    IBM Heron R2

    156physical qubits

    Updated Heron with lower crosstalk and larger circuit depth.

20233 events

  1. HardwareSuperconductingLane record

    IBM Condor

    1,121physical qubits

    IBM's first >1000-qubit processor.

  2. HardwareSuperconducting

    IBM Heron R1

    133physical qubits

    3–5× lower error rate than Eagle; the utility-scale successor.

  3. HardwareSuperconducting

    IBM Heron

    133physical qubits

20223 events

  1. Quantum volumeSuperconducting

    IBM Falcon r10 "Prague"

    512quantum volume

  2. Quantum volumeSuperconducting

    IBM Falcon r10 "Prague"

    256quantum volume

20211 event

  1. HardwareSuperconductingLane record

    IBM Eagle

    127physical qubits

    First >100-qubit superconducting processor.

20204 events

  1. Quantum volumeSuperconducting

    IBM Falcon r4 "Montreal"

    128quantum volume

  2. Quantum volumeSuperconducting

    IBM Falcon r4 "Montreal"

    64quantum volume

  3. Quantum volumeSuperconductingNew record

    IBM Raleigh

    32quantum volume

20198 events

  1. HardwareSuperconducting

    IBM Armonk

    1physical qubit

  2. HardwareSuperconducting

    IBM Q 53

    53physical qubits

  3. HardwareSuperconducting

    IBM Burlington

    5physical qubits

  4. HardwareSuperconducting

    IBM Essex

    5physical qubits

  5. HardwareSuperconducting

    IBM London

    5physical qubits

  6. HardwareSuperconducting

    IBM Ourense

    5physical qubits

Show older · 6 more

Sources

What the units count

Physical qubits
The hardware qubits on the device, before any error correction.
Atoms
Neutral-atom vendors count the atoms trapped in the array; each one can act as a physical qubit.
Algorithmic qubits (AQ)
IonQ's own system-level metric: the widest circuit at which a fixed set of benchmark algorithms still succeeds. No other vendor publishes it.
Logical qubits
Error-corrected qubits, each encoded across many physical ones. The codes differ between demonstrations, so two records rarely mean the same thing.
Quantum volume
A system-level score (a power of two) from square random circuits. Vendors run it under their own assumptions.
Two-qubit gate fidelity
How often a two-qubit gate does what it should, as a percent. The cards show the best published figure, with its device and source.

← All vendorsCompare IBM with other vendors