Skip to content
FusionBenchmark

Leaderboards

Devices ranked by what they have done. Two exclusions decide almost everything on this page, and both remove entries a careless table would show.

32 targets excluded

planned figures are permanently ineligible. A roadmap number is not a result.

157 design parameters excluded

What a machine was built to do, not what it has done. Without this, ITER tops “longest plasma” at 3,000 seconds having never operated.

Fusion triple product (nTτ_E)

Density × ion temperature × energy confinement time. The fairest single comparator across confinement approaches.

Ranked in 10²¹·m⁻³·keV·s. 2 targets excluded.

Fusion triple product (nTτ_E) ranking
#Device OperatorValue As publishedStatus
1 Tokamak Fusion Test Reactor Princeton Plasma Physics Laboratory 0.79 10²¹·m⁻³·keV·s 7.9 × 10²⁰ m⁻³·keV·s claimed 1995-02-17
2 Joint European Torus UK Atomic Energy Authority 0.61 10²¹·m⁻³·keV·s 6.1 × 10²⁰ m⁻³·keV·s claimed 1991-11-02
3 JT-60U National Institutes for Quantum and Radiological Science and Technology 0.56 10²¹·m⁻³·keV·s 5.6 × 10²⁰ m⁻³·keV·s claimed 1998
4 DIII-D General Atomics 0.37 10²¹·m⁻³·keV·s 3.7 × 10²⁰ m⁻³·keV·s claimed 1997
5 Alcator C 0.12 10²¹·m⁻³·keV·s 1.2 × 10²⁰ m⁻³·keV·s claimed 1984
6 C-Mod 0.074 10²¹·m⁻³·keV·s 7.4 × 10¹⁹ m⁻³·keV·s claimed 2016
7 Wendelstein 7-X Max Planck Institute for Plasma Physics 0.0517 10²¹·m⁻³·keV·s 6 × 10²⁶ K·m⁻³·s claimed 2018-06
8 Alcator A 0.024 10²¹·m⁻³·keV·s 2.4 × 10¹⁹ m⁻³·keV·s claimed 1978
9 ASDEX UPGRADE Max Planck Institute for Plasma Physics 0.022 10²¹·m⁻³·keV·s 2.2 × 10¹⁹ m⁻³·keV·s claimed 2016
10 Experimental Advanced Superconducting Tokamak Institute of Plasma Physics, Chinese Academy of Sciences 0.0096 10²¹·m⁻³·keV·s 9.6 × 10¹⁸ m⁻³·keV·s claimed 2015
11 KSTAR Korea Institute of Fusion Energy 0.0096 10²¹·m⁻³·keV·s 9.6 × 10¹⁸ m⁻³·keV·s claimed 2014
12 ASDEX 0.0072 10²¹·m⁻³·keV·s 7.2 × 10¹⁸ m⁻³·keV·s claimed 1988
13 ST40 Tokamak Energy 0.0063 10²¹·m⁻³·keV·s 6.3 × 10¹⁸ m⁻³·keV·s claimed 2022
14 TFR 0.0052 10²¹·m⁻³·keV·s 5.2 × 10¹⁸ m⁻³·keV·s claimed 1981
15 NSTX 0.0048 10²¹·m⁻³·keV·s 4.8 × 10¹⁸ m⁻³·keV·s claimed 2009
16 MAST 0.0045 10²¹·m⁻³·keV·s 4.5 × 10¹⁸ m⁻³·keV·s claimed 2006
17 PLT 0.0032 10²¹·m⁻³·keV·s 3.2 × 10¹⁸ m⁻³·keV·s claimed 1976
18 GLOBUS-M2 Ioffe Institute 0.0014 10²¹·m⁻³·keV·s 1.4 × 10¹⁸ m⁻³·keV·s claimed 2019
19 ST 2.4 × 10⁻⁴ 10²¹·m⁻³·keV·s 2.4 × 10¹⁷ m⁻³·keV·s claimed 1972
20 START 6.1 × 10⁻⁵ 10²¹·m⁻³·keV·s 6.1 × 10¹⁶ m⁻³·keV·s claimed 1998
21 T-3 2 × 10⁻⁵ 10²¹·m⁻³·keV·s 2 × 10¹⁶ m⁻³·keV·s claimed 1969

Ion temperature

Record with qualifier peak or sustained — they are different achievements and conflating them is a listed red flag.

Ranked in keV. 2 targets excluded.

Ion temperature ranking
#Device OperatorValue As publishedStatus
1 Tokamak Fusion Test Reactor Princeton Plasma Physics Laboratory 43.95 keV 5.1 × 10⁸ K claimed
2 JT-60U National Institutes for Quantum and Radiological Science and Technology 21.5 keV claimed
3 Joint European Torus UK Atomic Energy Authority 18.6 keV claimed 1991-11-02
4 DIII-D General Atomics 18.1 keV claimed 1997
5 ST40 Tokamak Energy 9.6 keV claimed 2022
6 KSTAR Korea Institute of Fusion Energy 8.6 keV claimed 2024-03
7 ASDEX UPGRADE Max Planck Institute for Plasma Physics 8 keV claimed 2016
8 Wendelstein 7-X Max Planck Institute for Plasma Physics 3.447 keV 4 × 10⁷ K claimed 2018-06
9 MAST 3 keV claimed 2006
10 C-Mod 2.5 keV claimed 2016
11 Experimental Advanced Superconducting Tokamak Institute of Plasma Physics, Chinese Academy of Sciences 2.1 keV claimed 2015
12 PLT 1.54 keV claimed 1976
13 Alcator C 1.5 keV claimed 1984
14 NSTX 1.2 keV claimed 2009
15 GLOBUS-M2 Ioffe Institute 1.2 keV claimed 2019
16 TFR 0.95 keV claimed 1981
17 Alcator A 0.8 keV claimed 1978
18 ASDEX 0.8 keV claimed 1988
19 ST 0.4 keV claimed 1972
20 T-3 0.3 keV claimed 1969
21 START 0.2 keV claimed 1998

Energy confinement time (τ_E)

How long the plasma retains its energy (τ_E) — stored energy divided by heating power. NOT the length of the discharge: a machine can hold a plasma for 1,000 seconds with a confinement time of a fraction of a second, and conflating the two is one of the most common errors in coverage.

Ranked in s. 2 targets excluded.

Energy confinement time (τ_E) ranking
#Device OperatorValue As publishedStatus
1 Joint European Torus UK Atomic Energy Authority 0.8 s claimed 1991-11-02
2 JT-60U National Institutes for Quantum and Radiological Science and Technology 0.69 s claimed 1998
3 Tokamak Fusion Test Reactor Princeton Plasma Physics Laboratory 0.28 s claimed 1995-02-17
4 DIII-D General Atomics 0.24 s claimed 1997
5 ASDEX 0.12 s claimed 1988
6 KSTAR Korea Institute of Fusion Energy 0.1 s claimed 2014
7 NSTX 0.08 s claimed 2009
8 ASDEX UPGRADE Max Planck Institute for Plasma Physics 0.056 s claimed 2016
9 Experimental Advanced Superconducting Tokamak Institute of Plasma Physics, Chinese Academy of Sciences 0.054 s claimed 2015
10 C-Mod 0.054 s claimed 2016
11 Alcator C 0.052 s claimed 1984
12 MAST 0.05 s claimed 2006
13 PLT 0.04 s claimed 1976
14 TFR 0.034 s claimed 1981
15 Alcator A 0.02 s claimed 1978
16 ST40 Tokamak Energy 0.012 s claimed 2022
17 ST 0.01 s claimed 1972
18 GLOBUS-M2 Ioffe Institute 0.01 s claimed 2019
19 T-3 0.003 s claimed 1969
20 START 0.003 s claimed 1998

Plasma duration

Wall-clock length of a sustained discharge. Distinct from confinement time, and routinely confused with it in press coverage.

Ranked in s. 39 design parameters excluded. 5 targets excluded.

Plasma duration ranking
#Device OperatorValue As publishedStatus
1 Experimental Advanced Superconducting Tokamak Institute of Plasma Physics, Chinese Academy of Sciences 1066 s claimed 2025-01-20
2 Wendelstein 7-X Max Planck Institute for Plasma Physics 480 s claimed 2023-02
3 KSTAR Korea Institute of Fusion Energy 102 s claimed 2026-06

Why some metrics have no leaderboard

Values are converted to each metric's canonical unit before sorting, and the figure as published is shown beside it. Ion temperature is held in kelvin for some devices and keV for others; ranking the raw numbers puts a 100 million degree plasma below a 40 million degree one. Where no conversion is known, the measurement is excluded rather than ranked on a number that means something else.