The Most Extreme Planets Ever Discovered: Heat, Speed, Diamonds, and Glass Rain

The most extreme planets ever discovered make our solar system look mild. KELT-9b has a dayside temperature of about 4,600 K (roughly 4,300°C), which the discovery team compared to a K4-type star. WASP-76b may rain iron on its night side. One planet candidate circles its star every 4.2 hours, and another is spiraling toward its sun. This article walks through the records, how each was measured, and which claims are solid versus still debated.

What counts as an exoplanet, and why the extremes are what we see first

An exoplanet is a planet orbiting a star other than the Sun. The first one found around a Sun-like star, 51 Pegasi b, was announced in 1995 by Michel Mayor and Didier Queloz, who shared the 2019 Nobel Prize in Physics for related work. NASA's Exoplanet Archive listed 6,451 confirmed planets when we last checked its count page: 4,785 found by transits, 1,204 by radial velocity, 292 by microlensing, and 98 by direct imaging. NASA announced that the tally had passed 6,000 in September 2025.

Those numbers are not a fair sample of the galaxy. Think of fishing with a net that has large holes: you mostly catch big fish, and you catch them in the stretch of water where you happen to cast. The transit method, which watches for a star to dim as a planet crosses its face, favors big planets on short orbits because they block more light and cross more often. The radial velocity method, which looks for a tiny back-and-forth Doppler wobble in starlight, favors heavy planets close to their stars for the same reason. Our guide to how astronomers find exoplanets covers both in detail.

The result is that the record books are crowded with hot Jupiters (gas giants that orbit in days instead of years) and with ultra-short-period rocky worlds. These are the easiest planets to detect, not necessarily the most common ones. A record in this field therefore describes the extreme end of what current instruments can see.

KELT-9b: the hottest planet known

KELT-9b was announced in 2017 by a team led by B. Scott Gaudi, based on observations from the Kilodegree Extremely Little Telescope survey. Its star, KELT-9, sits at the dividing line between spectral types A and B, with a surface temperature of about 10,170 K. That is nearly twice the Sun's surface temperature. For a refresher on how astronomers sort stars by temperature, see our explainer on stellar classification.

The planet circles KELT-9 every 1.48 days, and the discovery paper measured its dayside at about 4,600 K, "as hot as stars of stellar type K4." That is hotter than the surface of any red dwarf, the coolest and most common class of star, so KELT-9b's day side is hotter than many real stars, though not hotter than the Sun. Published estimates of the planet's mass vary with the analysis, from roughly 2 to 3 times Jupiter's.

The surprising fact: At about 4,600 K, KELT-9b's dayside is hot enough that most molecules cannot survive there. Gaudi's team argued that atomic metals, not molecules, must dominate the way its atmosphere absorbs light.

What the heat does to the atmosphere

At those temperatures, molecular hydrogen (H₂) splits into single hydrogen atoms on the dayside. Winds carry the atoms toward the cooler night side, where they recombine into molecules and release the energy they absorbed. A 2022 analysis in Astronomy & Astrophysics estimated that this cycle moves a meaningful share of heat from day to night, much as a refrigerant carries heat when it evaporates in one place and condenses in another.

The same heat ionizes metals. Astronomers have detected iron and singly ionized titanium in KELT-9b's atmosphere, and the 2023 detection of singly ionized terbium was the first for any exoplanet. These detections come from spectroscopy, the practice of splitting light into its component wavelengths to read chemical fingerprints. Our guide to spectroscopy explains the method.

KELT-9b is also being stripped. The discovery paper calculated that the planet receives about 700 times more extreme ultraviolet radiation than WASP-33b, another ultra-hot Jupiter, enough to remove much of its envelope over the star's main-sequence lifetime. A 2020 measurement put the mass loss at roughly 18 to 68 Earth masses per billion years. That is a slow leak next to the planet's size, but over the star's life it matters.

For comparison, the temperature scale runs far higher in other settings, from particle colliders to neutron-star mergers. Our piece on the most extreme temperatures in the universe puts KELT-9b in that context. A planet that hot is remarkable, but it is not close to the top of the full range.

The records at a glance

The table below lists the extremes covered in this article. Each row carries the year of the reported result and the primary source. Where two teams disagree, the text below explains how.

Record	Planet	Reported value	Year	Source
Hottest dayside	KELT-9b	About 4,600 K	2017	Gaudi et al., Nature
Fast winds measured by Doppler shift	HD 189733b	About 8,700 km/h	2015	Louden and Wheatley, University of Warwick release
Iron vapor condensing at night	WASP-76b	Day above 2,400°C; night about 1,500°C	2020	Ehrenreich et al., ESO release
Shortest-orbit candidate	KOI-1843.03	4.245-hour period	2013	Rappaport et al., ApJ Letters
Measured orbital decay	WASP-12b	Period shrinking 29 ms per year	2020	Yee et al., ApJ Letters
Extreme eccentricity	HD 80606b	Eccentricity about 0.93	2009	Laughlin et al., Nature
Lowest densities	Kepler-51 planets	Below 0.1 g/cm³	2020	Libby-Roberts et al., Astronomical Journal
Cool planet via microlensing	OGLE-2005-BLG-390Lb	About 5.5 Earth masses; about 50 K estimated	2006	Beaulieu et al., Nature

Several of these labels are narrower than they look. "Shortest orbit" refers to a candidate, not a fully confirmed planet. "Fast winds" refers to one measurement technique. None of these entries is a permanent champion, because new surveys keep adding rivals.

HD 189733b: the planet that looks blue

HD 189733b lies about 64 light-years away, close enough that astronomers have studied it more thoroughly than almost any other hot Jupiter. Bouchy and colleagues announced it in October 2005. It is slightly larger and heavier than Jupiter, and it orbits its star in about 2.2 days at roughly 0.031 AU, about one-twelfth of Mercury's distance from the Sun. Temperature estimates run from about 1,200 K to 1,500 K depending on the study.

In 2013, a team led by Thomas Evans used Hubble's STIS instrument to measure how much visible light the planet reflects, and found it to be a deep blue. That color is a clue to the atmosphere, not a sign of oceans. Small particles high in the atmosphere scatter blue light more than red, in the same way molecules in Earth's air tint our sky, and the best candidate for those particles is a haze of silicate (the family of minerals that includes glass). Earlier work by Lecavelier des Etangs and colleagues had pointed to magnesium silicate particles.

The 8,700 km/h wind

In 2015, a University of Warwick team led by Tom Louden used the HARPS spectrograph in Chile to watch the planet cross its star. The Doppler shift of sodium absorption lines let them measure motion in the atmosphere. After correcting for the planet's spin, they reported a wind of about 8,700 km/h (5,400 mph) blowing from the day side to the night side. That is more than seven times the speed of sound at Earth's surface.

The famous phrase "glass rain falling sideways" is a modeling inference, not an observation of drops. The logic runs like this: silicate vapor on the hot day side is carried toward the night side, where cooler temperatures let it condense, and fast winds would push the resulting grains horizontally. Telescopes cannot resolve raindrops on a planet 64 light-years away. What they measure are spectra, and the rain picture is the explanation that fits them best.

Hydrogen sulfide, and what it hints at

In July 2024, a team led by Guangwei Fu reported in Nature that JWST had detected hydrogen sulfide in HD 189733b's atmosphere, the first such detection on any exoplanet. The same data showed water, carbon dioxide, and carbon monoxide. The team estimated the planet's metallicity (the share of elements heavier than hydrogen and helium) at a few times the Sun's, and argued that the sulfur supports a formation history in which the planet gathered up many small solid bodies. Our overview of what JWST has found since 2022 places this result among the telescope's other atmosphere detections.

WASP-76b: where iron may rain

WASP-76b is about 640 light-years away. It is an ultra-hot Jupiter whose day side exceeds 2,400°C, hot enough to vaporize metals. In March 2020, ESO announced results from the ESPRESSO spectrograph on the Very Large Telescope. A team led by David Ehrenreich saw iron vapor on the evening edge of the planet, where day turns to night, but not on the morning edge.

The reading is that winds carry iron vapor from the day side to the night side, where the temperature falls to about 1,500°C. Iron condenses into droplets and leaves the visible atmosphere, so by the time the planet rotates back toward dawn the vapor is gone. The team titled the paper "Nightside condensation of iron in an ultrahot giant exoplanet."

An analogy helps. Picture a pot of boiling water with a cold lid: steam rises from the hot side and condenses on the cold surface. WASP-76b works like that pot, except the steam is iron and the pot is tidally locked, with one hemisphere always facing its star. Tidal locking means the planet's rotation period equals its orbital period, the same way the Moon always shows Earth one face.

This is a case where an extreme world teaches ordinary physics. Condensation of metals on a night side is the same process that forms clouds on Earth, run at a temperature where iron plays the role of water.

55 Cancri e: the diamond planet that may not be

55 Cancri e is a super-Earth (a rocky planet bigger than Earth but smaller than Neptune) about 1.86 times Earth's radius and roughly 7.7 times Earth's mass, according to a 2019 analysis cited in the planet's reference summary. It orbits its star every 0.7365 days, or about 17.7 hours, at a distance of about 0.016 AU. The star lies roughly 40 light-years away.

Astronomers first spotted it in 2004 by radial velocity. Early analysis got the period wrong, though: a 2010 study showed that the first signal was an alias, a false echo created by the gaps in observing times, and the true period was much shorter. Transits were announced in 2011, which let astronomers measure its size and density.

Where the diamond idea came from

In 2012, Nikku Madhusudhan and colleagues proposed that the planet might have a carbon-rich interior, with roughly a third of its mass in carbon and much of that in the form of diamond. The paper was titled "A possible carbon-rich interior in super-Earth 55 Cancri e," and the word "possible" is doing real work. It was a model fit to a measured radius and mass, and a rocky planet with iron, silicates, and an ordinary composition can also fit those numbers. The press coverage of a "diamond planet" outran what the model claimed.

What JWST found

Temperature measurements have been just as unsettled. Spitzer observations reported by Demory and colleagues in 2016 gave an average dayside near 2,700 K and a nightside near 1,380 K. A 2022 reanalysis found a dayside near 3,771 K. The thermal emission also seems to vary over time, and volcanic activity has been proposed as one cause.

In May 2024, Renyu Hu and colleagues published JWST results in Nature. Using NIRCam and MIRI, they measured a thermal emission spectrum from 4 to 12 micrometers and concluded the planet cannot be a bare lava world wrapped in a thin atmosphere of vaporized rock. Instead, they reported a volatile atmosphere likely rich in carbon dioxide or carbon monoxide, possibly sustained by outgassing from a magma ocean. That finding does not rule out diamond deep inside, but it shows the planet is more complicated than the nickname suggests.

KOI-1843.03 and WASP-12b: orbits that end badly

The shortest orbital periods known belong to planets and candidates that skim the surfaces of their stars. KOI-1843.03, found in data from NASA's Kepler mission, has an orbital period of 4.245 hours. In 2013, Saul Rappaport and colleagues estimated its radius at about 0.6 Earth radii and used a tidal argument to put a floor on its density.

The argument is based on the Roche limit, the distance inside which a star's tidal pull exceeds the self-gravity holding a planet together. A loose pile of rubble would be torn into a ring there. A planet can survive only if it is dense enough, and for KOI-1843.03 the team calculated a mean density above about 7 g/cm³. That points to a mostly iron body, with at most about 30 percent silicates by mass. For comparison, Earth's mean density is about 5.5 g/cm³, so this planet would be markedly denser than our own.

A 2019 reanalysis by Price and Rogers used three-dimensional models and argued that the planet must be at least about 60 percent iron by mass, with a mass between roughly 0.32 and 1.06 Earth masses. The conclusion is similar, and the objects are exotic either way. They may be the stripped cores of larger planets whose outer layers were boiled away. Note that KOI-1843.03 is a Kepler candidate, so "shortest orbit" is a claim about a candidate, and new discoveries can displace it.

WASP-12b is spiraling inward

WASP-12b is a hot Jupiter with a 1.09-day orbit around a late-F star. In 2020, Samuel Yee and colleagues reported that the interval between its transits has been shrinking by 29 ± 2 milliseconds per year since its discovery in 2008. They ruled out a light-travel-time explanation using new radial velocity data, and called it the first planetary system in which orbital decay can be stated with confidence.

The decay timescale, defined as the period divided by its rate of change, is 3.25 ± 0.23 million years. That is not a countdown to a date. It is a rough scale for how long the process can run before the planet is shredded or swallowed. In human terms the change is minuscule, but astronomers have timed transits for more than a decade, and a few milliseconds a year adds up.

Why does it happen? The planet's gravity raises a tidal bulge on the star, and because the planet orbits faster than the star rotates, the bulge drags on the planet and drains orbital energy. Our explainer on planetary migration covers the broader family of processes that move planets around.

HD 80606b: a hot Jupiter on a comet orbit

HD 80606b was announced in April 2001 by the Geneva Extrasolar Planet Search team led by Michel Mayor and Didier Queloz. It is about four times Jupiter's mass and orbits a star roughly 217 light-years away. Its orbit takes 111.4 days, and its eccentricity is about 0.93, where 0 is a circle and 1 is an escape path.

That eccentricity means the planet swings from about 0.89 AU at its farthest point (close to the distance of Venus from the Sun) to about 0.031 AU at its closest, a gap of roughly 29 to 1. The Spitzer Space Telescope watched one close pass in 2009. A team led by Gregory Laughlin reported that the planet's 8-micron brightness temperature rose from about 800 K to 1,500 K in roughly six hours as it received its strongest irradiation.

The speed of the response is the interesting part. The team estimated a radiative time constant of about 4.5 hours at the layer they observed, compared with several days in Earth's stratosphere. In other words, that layer can heat and cool far faster than air on Earth, because it sits in a thin, hot, dry atmosphere with nothing like the heat-storing oceans that smooth out our weather.

Kepler-51: planets with the density of foam

Not all extremes involve heat. Three planets around the star Kepler-51 have masses only a few times Earth's but radii larger than Neptune's, which puts them in a class called super-puffs. In 2020, Libby-Roberts and colleagues combined new Hubble transit timings with reanalyzed Kepler data and confirmed that all three have densities below 0.1 g/cm³. Water is 1 g/cm³, so these planets are less than a tenth as dense as water, and the "cotton candy planets" nickname is press shorthand.

The same team used Hubble to measure Kepler-51b and Kepler-51d's transmission spectra, which turned out featureless between 1.15 and 1.63 micrometers. That rules out strong water absorption and points to a high-altitude layer of haze or cloud that hides the gas below. A 2025 JWST study of Kepler-51d reported the thickest haze found on a planet yet, according to coverage of that work.

How such low-density worlds form and survive is an open question. A light, puffy atmosphere on a small core should leak away quickly when heated by its star. Whatever allows them to persist, it is not yet pinned down.

Rogue planets: worlds with no star

A rogue planet, formally a free-floating planetary-mass object, travels through space without orbiting a star. Two formation routes are known. A planet can be flung out of its home system by the gravity of a heavier neighbor, or a small clump of gas can collapse on its own without ever becoming massive enough to ignite fusion.

How many exist is a story of correction. A 2011 microlensing analysis by Sumi and colleagues found an excess of very short events, which they read as evidence for Jupiter-mass free-floating planets almost twice as common as main-sequence stars. Headlines turned that into "100 billion rogue planets." A 2017 study by Przemek Mróz and colleagues analyzed a sample six times larger, from the OGLE survey, found no such excess, and set a 95 percent upper limit of 0.25 Jupiter-mass free-floating or wide-orbit planets per main-sequence star.

The 2017 result concerns Jupiter-mass objects. The same data suggest that lower-mass free-floating planets, from Earth to super-Earth size, could be common, though those counts are less certain. Microlensing, the brief brightening of a background star when a foreground mass bends its light, is the main detection tool, and each event is a one-time alignment that cannot be repeated.

NASA says its Nancy Grace Roman Space Telescope will detect hundreds of rogue planets in its Galactic Bulge Time-Domain Survey. Check NASA's mission page for the current launch and survey status. If you want a scenario rather than a survey, our thought experiment on a rogue planet entering our solar system explores what such an encounter might do.

What extremes tell us about how planets form

Each oddity is a puzzle, and together they form a test of planet-formation theory. The standard picture says giant planets build up far from their star, beyond the frost line, the distance where water and other volatile compounds freeze into solid ice grains. Past that line there is more solid material to build a massive core. Yet hot Jupiters orbit well inside the equivalent of Mercury's orbit.

The proposed explanation is migration. In a 1996 paper, Lin, Bodenheimer, and Richardson showed how a giant planet's interaction with its birth disk could pull it inward to the position of 51 Pegasi b. Other routes, including gravitational scattering and slow tidal circularization of highly elliptical orbits, are also considered, and HD 80606b may be caught partway through the latter. Astronomers still debate how much each mechanism contributes.

The practical lesson is that our solar system may be one outcome among several. If hot Jupiters are rare in absolute terms (they are detected often because they are easy to detect, not because they are numerous), then a system with a Jupiter-like planet far from the star and rocky planets close to it is not an anomaly. Still, the question of how typical our arrangement is remains open, and the extremes are some of the best evidence for that uncertainty.

What people get wrong about extreme planets

"KELT-9b is hotter than the Sun." It is not. The Sun's surface is about 5,800 K, and KELT-9b's dayside is about 4,600 K. The planet is hotter than the coolest stars, and its host star is far hotter than the Sun.

"HD 189733b is an ocean world because it is blue." The color comes from scattering by haze, which is a physical mechanism unrelated to water. Conditions there are hostile to any known biochemistry.

"55 Cancri e is a ball of diamond." The 2012 model said a large fraction of the planet might be carbon, much of it possibly diamond. JWST found a volatile atmosphere in 2024. The interior is not directly observable.

"Record" means permanent. Most records here are one detection method's best result. A rocky planet orbiting in 4.2 hours may be displaced by a new candidate next year, and a hotter gas giant may turn up in a survey that has not yet run. Treat the records as a snapshot.

"Rogue planets outnumber stars." For Jupiter-mass objects, the 2017 result puts the upper limit at 0.25 per star. The question remains open for smaller worlds.

Frequently asked questions

Q: What is the hottest planet ever discovered? A: KELT-9b has the highest measured dayside temperature, about 4,600 K according to the 2017 discovery paper. Its host star is an A/B-type star near 10,170 K, and the planet orbits it every 1.48 days. Later analyses give similar values, though uncertainties of a few hundred kelvin are normal for this kind of measurement.

Q: Is HD 189733b really blue like Earth? A: It looks blue in Hubble measurements from 2013, but for a different reason. Particles high in the atmosphere, likely silicate haze, scatter blue light, and the planet has no oceans. Temperatures run about 1,200 K to 1,500 K depending on the study, and winds of about 8,700 km/h were measured in 2015.

Q: Is 55 Cancri e actually made of diamond? A: Nobody knows. A 2012 model said the interior might be carbon-rich, with much of that carbon as diamond, but the paper called it only a possibility. JWST observations in 2024 found a likely carbon dioxide or carbon monoxide atmosphere, which does not settle the question of the interior.

Q: How can a planet survive an orbit of only a few hours? A: It needs to be dense enough to resist the star's tides. For KOI-1843.03, with a 4.245-hour orbit, the 2013 analysis calculated a minimum mean density near 7 g/cm³, which implies a largely iron composition. A less dense body at that distance would be pulled apart, so these objects are probably dense cores.

Q: What is a rogue planet and how do astronomers find one? A: A rogue planet travels through space without a host star. Astronomers find them by microlensing, when the planet's gravity briefly magnifies a distant background star. A 2017 survey set an upper limit of 0.25 Jupiter-mass free-floating planets per star, much lower than a 2011 estimate had suggested.

Q: Could life exist on an ultra-hot Jupiter like KELT-9b? A: No known form of life could survive there. At about 4,600 K, molecules break apart on the day side, and the planet has no solid surface. Astronomers study such planets as physics laboratories, since they test models of atmospheres under conditions no Earth laboratory can reproduce.

Q: Why are so many extreme planets hot Jupiters? A: Because the methods that find planets are biased toward big, close-in worlds. Transits and radial velocity both detect those planets most easily. The record list reflects detectability as much as what is common in the galaxy.

The takeaway

The planets in this article are less a gallery of freaks than a set of experiments nature ran for us. KELT-9b shows what an atmosphere does when hydrogen molecules cannot survive, and WASP-76b shows iron behaving like water in a weather cycle. KOI-1843.03 and WASP-12b show how tides set the limits on how close a planet can live, and Kepler-51 shows that a planet can be lighter than we expected.

Just as important, several famous claims have been revised: the "100 billion rogue planets" figure shrank in 2017, the "diamond planet" met a volatile atmosphere in 2024, and "glass rain" remains an inference. When you read the next planetary record, check whether it is a measurement, a model, or a candidate. The answer changes how much weight it should carry.

For more on how astronomers read these worlds, see our guide to how exoplanet atmospheres are detected. For scale, compare the numbers here with the most massive objects in the universe and the neutron stars that sit at the extreme end of stellar physics.

Sources

• Gaudi, B.S. et al. (2017), "A giant planet undergoing extreme-ultraviolet irradiation by its hot massive-star host," Nature 546: arXiv:1706.06723
• Rappaport, S. et al. (2013), "The Roche limit for close-orbiting planets: Minimum density, composition constraints, and application to the 4.2-hour planet KOI 1843.03": arXiv:1307.4080
• Mróz, P. et al. (2017), "No large population of unbound or wide-orbit Jupiter-mass planets": arXiv:1707.07634
• Yee, S.W. et al. (2020), "The Orbit of WASP-12b Is Decaying": arXiv:1911.09131
• Libby-Roberts, J.E. et al. (2020), "The Featureless Transmission Spectra of Two Super-Puff Planets": arXiv:1910.12988
• ESO (2020), "ESO's VLT observes exoplanet where it rains iron": eso2005
• NASA Exoplanet Archive, confirmed planet counts
• Reference summaries consulted for orbital parameters: KELT-9b, HD 189733 b, 55 Cancri e, HD 80606 b