Fast Blue Optical Transients (FBOTs)

• FBOTs are rapidly evolving, blue, luminous extragalactic transients defined by rapid rise, short duration, high temperatures, and diverse powering mechanisms.
• Observations show that FBOTs share spectral overlaps with interacting supernovae, implying multiple progenitor channels and dense circumstellar environments.
• Multiwavelength studies revealing radio and X-ray emissions have been key to distinguishing between dense CSM interaction and central engine models in FBOT events.

Fast blue optical transients (FBOTs) are an observational class of rapidly evolving, blue, and luminous extragalactic transients. Across the literature, the term denotes phenomenology rather than a unique power source: the class is defined by fast optical rise and decline, hot blue continua, and high peak luminosity, while the physical interpretations range from dense circumstellar-medium (CSM) interaction to central-engine activity, compact-object accretion, jet/cocoon emission, and accretion-induced collapse scenarios. The most extreme AT2018cow-like events add unusually strong radio and X-ray emission to the optical phenomenology, but broader photometric samples show that the FBOT label also includes objects overlapping with more familiar interacting and stripped-envelope supernova populations (Ni et al., 1 Jul 2026, Ho et al., 2021).

1. Phenomenological definition and class boundaries

The class is operationally defined in several, not fully identical, ways. One widely used phenomenological description is “rapidly evolving, blue, and luminous events” (Ni et al., 1 Jul 2026). In the Dark Energy Survey (DES) host-galaxy study, rapidly evolving transients (RETs), treated there as FBOTs, were defined by light curves that rise to peak in less than about 10 days, decline to 10% of peak brightness within about 30 days, span a wide range of peak luminosities, and can be detected to redshift $z>1$ (Wiseman et al., 2020). A targeted FAST radio search described FBOTs as luminous, rapidly evolving events with blue spectra, durations above half-maximum brightness of $\lesssim 12$ days, peak optical luminosities up to $\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$, and effective temperatures around $T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$ (Gao et al., 25 Aug 2025).

Search pipelines formalize these ideas with survey-dependent cuts. In the ZTF Phase I search, the core temporal criterion was a $g$-band duration above half-maximum light

$1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$

and the paper used the term FBOT more narrowly for the blue subset with peak color $g-r\lesssim -0.2$ mag (Ho et al., 2021). In the DES sample, observer-frame selection used

$$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$

explicitly because many objects lacked redshifts at selection time (Wiseman et al., 2020).

A central classification point is that FBOT is not a single physical class. The ZTF search identified 38 candidate extragalactic fast transients, of which 28 were blue enough to be treated as FBOTs; among the 38, 19 were spectroscopically classified as core-collapse supernovae, including Type II/IIb/Ib, Type IIn/Ibn, and Type Ic/Ic-BL events (Ho et al., 2021). This directly argues against treating every fast blue transient as an AT2018cow analog. The AT2018cow-like subset is instead a rare, high-luminosity, multiwavelength-loud extreme within a broader phenomenological umbrella.

2. Optical phase space and continuity with interacting supernovae

A major recent development is the uniform comparison of FBOTs with SNe Ibn and SNe Icn, both of which are hydrogen-poor interacting explosions. A 25-object sample comprising 10 SNe Ibn, 5 SNe Icn, and 10 FBOTs was analyzed using Gaussian-process reconstructions of multiband optical light curves in a common $g$-band framework (Ni et al., 1 Jul 2026). The measured observables were the peak epoch $t_{\rm peak}$, the peak monochromatic luminosity $\lesssim 12$0, the rise timescale $\lesssim 12$1, and the post-peak decline proxy

$\lesssim 12$2

The rise timescale was defined as the interval between $\lesssim 12$3 and the pre-peak time when the flux reaches $\lesssim 12$4.

Two empirical planes were emphasized: $\lesssim 12$5 and

$\lesssim 12$6

In both planes, FBOTs preferentially occupy the luminous, rapidly evolving end of the distribution, but the classes are not cleanly separated. The FBOT versus Ibn+Icn overlap coefficient is low, $\lesssim 12$7 in the peak-luminosity–rise-time plane and $\lesssim 12$8 in the decline–rise-time plane, yet still nonzero (Ni et al., 1 Jul 2026). Peak colors, defined at $\lesssim 12$9-band maximum as $\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$0, overlap much more strongly: $\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$1, with FBOTs extending to somewhat bluer colors.

This optical result matters because it weakens a rigid categorical distinction between FBOTs and interacting stripped-envelope transients. It does not imply identity of mechanism for all objects. Rather, it shows that at least part of FBOT optical phenomenology inhabits the same empirical phase space as dense-CSM-powered SNe Ibn/Icn. A common misconception is that blue color and short timescale alone isolate an entirely separate explosion class; the uniform comparison indicates that the separation is incomplete in observed light-curve space (Ni et al., 1 Jul 2026).

3. Host galaxies, stellar populations, and progenitor environment

Host-galaxy studies place the broader RET/FBOT population in young stellar environments. In the DES sample of 106 RETs, 96/106 had a detected host in deep photometry, 49/106 had secure host-galaxy redshifts with stellar-mass and star-formation-rate estimates, and 37/106 had metallicity measurements (Wiseman et al., 2020). The principal environmental result is that these transients explode exclusively in star-forming galaxies. Mean host properties were reported as

$\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$2

$\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$3

and

$\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$4

Relative to comparison samples, RET/FBOT hosts prefer high specific star-formation rates and appear chemically under-enriched. The DES analysis found them similar to stripped-envelope supernova hosts, less extreme than long-duration gamma-ray burst and superluminous supernova hosts, and inconsistent with an old stellar-population origin (Wiseman et al., 2020). It also found no clear relation between host properties and transient peak magnitudes or decline rates.

AT2018cow-like events fit this young-host picture. AT2020xnd was found in a dwarf galaxy with estimated stellar mass roughly $\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$5 to $\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$6, likely around $\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$7, and $\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$8 (Perley et al., 2021). The same paper emphasized that the then-known AT2018cow-like radio-luminous sample occupied low-mass, star-forming dwarfs. A plausible implication is that the observational class contains multiple engine or interaction channels, but these channels predominantly trace massive-star evolution rather than long-delay thermonuclear populations.

4. Multiwavelength phenomenology: radio, X-rays, timing, and FRB constraints

Radio observations show that at least the luminous subset of FBOTs launches fast outflows into dense local environments. A forward-shock synchrotron model with a broken power-law, radially confined CSM reproduces the radio diversity of a representative FBOT sample, including very steep post-peak declines, with inferred shock velocities

$\gtrsim 10^{44}\,\mathrm{erg\,s^{-1}}$9

mass-loading rates

$T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$0

and total CSM masses

$T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$1

In this framework, early radio evolution is set by synchrotron self-absorption and sometimes external free-free absorption, while the steep late-time decline marks the forward shock crossing from a dense inner shell into a lower-density exterior (Liu et al., 17 May 2026).

Well-observed individual events illustrate the same pattern. AT2020xnd reached

$T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$2

at 20 GHz around 75 days, with X-ray luminosity peaking near

$T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$3

and radio modeling implied a trans-relativistic outflow with $T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$4 propagating into a dense CSM with effective $T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$5 for $T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$6 (Bright et al., 2021). AT2018cow low-frequency radio monitoring extended this picture to larger radii, finding $T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$7 up to 257 days and mass-loss rates at radii corresponding to 19.3–45.7 years before explosion that were about 100 times smaller than the rate inferred at 2.2 years before explosion, consistent with enhanced late-stage mass loss and a dense inner shell (Nayana et al., 2021).

The X-ray channel is often harder to reconcile with a single shock component. AT2022tsd reached an initial 0.3–10 keV luminosity of $T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$8 at $T_{\rm eff}\sim 2\times 10^4\,\mathrm{K}$9 d, with an absorbed power-law spectrum of average photon index $g$0, no statistical evidence for spectral evolution, and a light curve consistent either with $g$1 or with a broken power law with $g$2 d (Matthews et al., 2023). In AT2020xnd, the X-ray spectrum and luminosity were too bright and too hard to be the high-frequency continuation of the radio synchrotron component, implying a separate emission component, possibly from accretion onto a newly formed black hole or neutron star (Bright et al., 2021).

Timing adds another layer of complexity. AT2018cow showed a possible X-ray quasi-periodicity near 250 s, detected in XMM-Newton PN data at 99.76% significance and independently supported at the same frequency in stacked Swift data at 94.26%, suggesting a stable engine-related clock over $g$3 cycles (Zhang et al., 2022). That signal was discussed as compatible with an accreting compact object and, in one interpretation, a tidal disruption event by an intermediate-mass black hole.

FRB searches constrain, but do not eliminate, newborn-magnetar scenarios. Deep FAST observations of AT2018cow and CSS161010 detected no astrophysical single-pulse candidates, established flux-density limits near $g$4 mJy, and, assuming an FRB 121102-like log-normal luminosity function, constrained the intrinsic burst rate to

$g$5

For typical FBOT parameters, the free-free transparency time at 1.25 GHz was estimated as about 2.6 yr, shorter than the 4–6 yr post-explosion observing epochs, so the null result was argued to be unlikely to arise primarily from ejecta absorption (Gao et al., 25 Aug 2025).

5. Physical models and proposed power sources

The model space is plural. One prominent line of work treats at least some FBOTs as dense-CSM interaction events. In the unified TransFit-CSM framework, thin-shell ejecta–CSM interaction is coupled to time-dependent radiative diffusion through an optically thick CSM shell with wind-like density profile $g$6. The model includes shock heating plus an effective inner heating term, treated formally as $g$7Ni heating but explicitly interpreted as an effective inner power source rather than literal evidence for large nickel masses in every case. The radiative output is mapped to observables through

$g$8

In fitted parameter space, FBOTs overlap substantially with SNe Ibn/Icn, with reported FBOT versus Ibn+Icn overlap coefficients $g$9 in the $1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$0–$1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$1 plane and $1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$2 in the $1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$3–$1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$4 plane (Ni et al., 1 Jul 2026). This supports a common dense-CSM interaction framework for at least part of the class.

Binary-mediated CSM formation provides one route to such environments. Viscous circumbinary-disc models formed by rapid, stable mass transfer from expanding helium stars yield compact H-poor CSM with mass $1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$5–$1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$6, half-mass radius $1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$7–$1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$8, and aspect ratio $1\,\mathrm{d} < t_{1/2,g} < 12\,\mathrm{d},$9, sufficient to power the fastest interacting Type Ibc events and some FBOT-like transients such as SN 2018gep and SN 2019jc (Chiba et al., 21 Apr 2026).

A second major family invokes compact central engines. Magnetar-diffusion fits to 40 FBOTs reported median parameters

$g-r\lesssim -0.2$0

with spin-down luminosity

$g-r\lesssim -0.2$1

That study argued that FBOTs are the low-ejecta-mass end of a broader magnetar-powered sequence connecting SLSNe, SNe Ic-BL, and some lGRB-SNe, and emphasized an empirical ejecta-mass upper limit of about $g-r\lesssim -0.2$2 for FBOTs (Liu et al., 2022).

Jet/cocoon models instead place FBOTs within engine-driven core collapse. In that picture, a hydrogen-rich collapsing star launches a relativistic jet; jet–star interaction inflates a mildly relativistic inner cocoon of typical energy $g-r\lesssim -0.2$3 erg, whose cooling emission produces a rapid optical decline $g-r\lesssim -0.2$4, while CSM interaction generates the radio signal and the jet or cocoon becomes optically thin to X-rays on day timescales (Gottlieb et al., 2022). Time-dependent outflow and radiative reprocessing models adopt a different engine phenomenology: an optically thick outflow launched over a few days diffuses and reprocesses central energy into the observed UV-optical emission. Fits to selected FBOTs gave outflow masses $g-r\lesssim -0.2$5–$g-r\lesssim -0.2$6 and ejection durations of a few days (Chen et al., 2022), while a frequency-dependent reprocessing calculation predicted a non-blackbody NIR slope

$g-r\lesssim -0.2$7

and, when applied to AT2018cow, inferred $g-r\lesssim -0.2$8, favoring a black-hole remnant if the progenitor was a blue supergiant of $g-r\lesssim -0.2$9 (Chen et al., 2024).

A third group of models ties extreme FBOTs to compact-object accretion or collapse channels. AT2024wpp, described as the most luminous known FBOT, showed late-time optical flattening and X-ray rebrightening consistent with

$$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$0

and was modeled as a $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$1 Wolf-Rayet star merging with a $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$2 black hole, with fallback beginning at $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$3 days (Liu et al., 24 Feb 2026). AT2018cow has also been interpreted through a possible $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$4 s X-ray quasi-periodicity and TDE-like X-ray evolution as a candidate intermediate-mass-black-hole event (Zhang et al., 2022). At the same time, accretion-induced-collapse models of merged white dwarfs or late AIC of a super-Chandrasekhar double-WD remnant have been proposed to explain the small ejecta masses, engine-powered optical emission, and long-lived X-rays of FBOTs (Lyutikov et al., 2018, Lyutikov, 2022).

6. Rates, subclass structure, and open controversies

Population studies now separate the broad photometric FBOT/RET population from the rarer AT2018cow-like subset. The ZTF search concluded that the broader $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$5 d fast-transient population occurs at about 7% of the core-collapse supernova rate, with a 95% confidence interval of $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$6 to $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$7, but that AT2018cow-like events represent at most $$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$8–$$-1.5 < (g-r) < 0.6,\qquad t_{\mathrm{rise},r} < 16.5~\mathrm{days},$$9 of the local core-collapse supernova rate (Ho et al., 2021). This is consistent with the spectroscopic heterogeneity of fast blue samples and with the rarity of luminous radio/X-ray counterparts.

The main controversy is therefore not whether FBOTs exist observationally, but what the label means physically. One pole of the debate holds that at least some FBOTs are dense-CSM interaction-powered events lying on a continuum with SNe Ibn and SNe Icn; the other emphasizes that the most extreme members, especially AT2018cow-like objects with luminous X-ray or radio emission, require additional power sources or non-thermal components beyond optical CSM interaction models (Ni et al., 1 Jul 2026). A related misconception is that “FBOT” automatically implies a magnetar, a black hole, a jet, or a TDE. The current literature does not support any single one-to-one mapping.

Radio and high-energy observations are the principal discriminants. Radio light curves probe shock velocity, absorption, CSM density, and the radial extent of the immediate pre-explosion environment (Liu et al., 17 May 2026). X-rays diagnose whether a separate compact-engine component is present, and timing behavior such as the possible AT2018cow quasi-periodicity further tests accretion-based interpretations (Zhang et al., 2022). FRB nondetections already exclude the simplest picture in which FBOT-born magnetars behave like very active repeaters with FRB 121102-like luminosity statistics at the epochs and sensitivities observed, but they do not rule out magnetars in general (Gao et al., 25 Aug 2025).

The most robust synthesis is therefore conditional. FBOTs are best treated as an observational umbrella containing multiple channels. The class includes ordinary interacting and shock-cooling supernovae selected by fast, blue photometric criteria; a dense-CSM interaction sequence overlapping with SNe Ibn/Icn; and a much rarer luminous subset with strong evidence for central engines, compact-object accretion, or unusually structured CSM. This suggests that future progress depends less on refining a single universal definition than on uniform multiwavelength modeling, dense early-time sampling, and timely sustained radio and high-energy follow-up of newly discovered events.