RUBIES: Red Infrared Extragalactic Survey

• RUBIES is a JWST/NIRSpec multi-object spectroscopic survey of bright, red, high-z galaxies, using both low-resolution PRISM and medium-resolution G395M modes.
• The survey employs a tractable selection function and forward-modeling to robustly measure galaxy properties such as metallicity, star-formation histories, and quenching processes.
• RUBIES also serves as a calibration framework for photometric diagnostics and identifies diverse sources including Little Red Dots, AGN, early quiescent galaxies, and Galactic brown dwarfs.

RUBIES, the Red Unknowns: Bright Infrared Extragalactic Survey, is a JWST/NIRSpec multi-object spectroscopic program built to obtain a spectroscopic census of bright, red distant sources selected from deep NIRCam imaging in the EGS and UDS/PRIMER fields. In current usage, the name primarily denotes the extragalactic survey rather than the mineralogical term. Across the literature, RUBIES functions both as a survey infrastructure—combining NIRSpec/PRISM and G395M spectroscopy with deep JWST imaging—and as a framework for several distinct science programs, including the census of Little Red Dots, the confirmation of very early quiescent galaxies, studies of dusty star-forming galaxies, high-redshift metallicity measurements, and calibration of photometric diagnostics against spectroscopy (Hviding et al., 5 Jun 2025, Cooper et al., 2024).

1. Survey definition and observing architecture

RUBIES is a JWST Cycle 2 NIRSpec/MSA program, GO 4233, designed as a “complete census of the bright and red distant Universe with JWST/NIRSpec” and, in a complementary formulation, to obtain essentially complete spectroscopy for bright, red galaxies at and beyond cosmic noon (Hviding et al., 5 Jun 2025, Cooper et al., 2024). In one survey-level characterization, it targets ∼4500 galaxies over ∼150 arcmin² in UDS and EGS (Hviding et al., 5 Jun 2025).

The observational backbone is NIRSpec multi-object spectroscopy in two modes. PRISM/CLEAR supplies low-resolution, broad wavelength coverage across the full NIRSpec range, while G395M/F290LP supplies medium resolution over the redder part of the bandpass. For representative RUBIES observations, PRISM/CLEAR is described as $R\sim100$ over the full NIRSpec range and G395M/F290LP as $R\sim1000$ at 2.9–5.1 $\mu$m, with typical exposure times of 48 min per grating per object or mask (Cooper et al., 2024, Hviding et al., 5 Jun 2025). This pairing is central to the survey’s strategy: PRISM constrains continuum shape and broad spectral structure, while G395M resolves Balmer and forbidden-line kinematics.

RUBIES is embedded in the major JWST imaging fields rather than operating as a standalone photometric program. In EGS it interfaces naturally with CEERS imaging, while in UDS it interfaces with PRIMER. This architecture is what enables the recurrent survey pattern seen across the literature: NIRCam photometric preselection, NIRSpec spectroscopic confirmation, then joint forward modeling of continua, line profiles, stellar populations, AGN components, or synthetic observables depending on the science case (Morales et al., 3 Jul 2025, Turner et al., 19 Sep 2025).

2. Selection function and analysis framework

A defining feature of RUBIES is that its selection is simple enough to be modeled explicitly. One survey description states that targeting priority is set by F444W flux, F150W–F444W color, and photometric redshift, with weights proportional to the inverse number density in that three-dimensional observable space; bright, red, high-$z$ galaxies therefore receive higher weight (Lewis et al., 2 Dec 2025). Another describes the program as explicitly optimized for very red NIRCam sources, including a highest-priority class with F150W–F444W $>3$ and F444W $<27$ (Cooper et al., 2024). For the LRD census, the survey is further characterized as having no morphological pre-selection, which is important because it allows compact AGN-dominated systems, resolved galaxies, and intermediate cases to enter the same spectroscopic sample (Hviding et al., 5 Jun 2025).

This tractable selection function has methodological consequences. In the mass–metallicity analysis, the RUBIES selection function is forward-modeled in observable space and shown to have a negligible effect on the measured MZR relative to other biases, particularly non-Gaussian metallicity posteriors and line observability (Lewis et al., 2 Dec 2025). In the FLARES comparison for RUBIES-UDS-QG-z7, synthetic spectra and photometry are processed through the same NIRCam/NIRSpec filter set and comparable SED-fitting assumptions as the observed RUBIES object, so the simulation–data comparison is explicitly performed in observable space rather than via intrinsic simulation properties alone (Turner et al., 19 Sep 2025).

At the spectral-analysis level, RUBIES papers repeatedly emphasize joint treatment of PRISM and G395M. The broad-line census introduces unite, a fitting framework that simultaneously models PRISM and G395M line complexes with shared intrinsic source parameters and explicit handling of resolution, flux offsets, and line-spread functions; broad-line identification is then based on WAIC comparison between narrow-only and broad+narrow models (Hviding et al., 5 Jun 2025). Other RUBIES studies adopt Prospector for combined photometric–spectroscopic inference, non-parametric SFHs, flexible dust laws, and explicit AGN continuum components when needed (Wang et al., 2024, Cooper et al., 2024).

3. Little Red Dots, broad-line AGN, and black-hole activity

RUBIES is closely associated with the spectroscopic consolidation of the Little Red Dot population. In the survey-wide census of $z>3.1$ galaxies, a sample of 1482 galaxies yields 80 broad-line sources, with 28 (35%) at $z>6$, and from these a subpopulation of 36 spectroscopic LRDs is defined by the simultaneous presence of a broad Balmer line, a v-shaped UV-to-optical continuum, and a dominant rest-optical point source (Hviding et al., 5 Jun 2025). The most concise summary of the result is the paper’s main empirical statement: all v-shaped point sources have broad lines, to the extent allowed by the data (Hviding et al., 5 Jun 2025). The same work also shows that previously published photometric LRD searches, while accurate, recover only 50–62% of the bright RUBIES LRDs at $\mathrm{F444W}<26.5$, with the missed fraction driven by faint rest-UV photometry, comparatively blue rest-optical colors, or highly uncertain photometric redshifts (Hviding et al., 5 Jun 2025).

RUBIES has also supplied some of the sharpest physical case studies within the LRD class. “The Cliff,” a RUBIES LRD at $z=3.55$, is reported from new NIRSpec/IFU data to have low metallicity $R\sim1000$0 inferred from the low narrow-line [OIII]$R\sim1000$1/H$R\sim1000$2 ratio together with the non-detection of [OII] and [NII]; at the same time it hosts an overmassive black hole relative to the local $R\sim1000$3–$R\sim1000$4 expectation (Ivey et al., 10 Apr 2026). That paper argues that simulation runs capable of reproducing the system require high seed masses of $R\sim1000$5, and that such runs are as rare in the simulation volume as The Cliff is in the RUBIES volume over $R\sim1000$6 (Ivey et al., 10 Apr 2026).

RUBIES has further been central to the emerging view that Balmer-line physics in LRDs is not always described by simple Case B recombination. In a dedicated analysis of seven high-S/N LRDs, one of the two RUBIES objects, RUBIES-EGS-49140, is found to depart from Case B expectations in its broad Balmer ratios by more than $R\sim1000$7 (Nikopoulos et al., 7 Oct 2025). The same paper concludes that the anomaly is consistent with high-density, optically thick gas surrounding the black hole, and that for LRDs more generally, deriving extinction or black-hole masses from H$R\sim1000$8/H$R\sim1000$9 alone can be misleading unless higher-order Balmer lines and broad/narrow decomposition are also available (Nikopoulos et al., 7 Oct 2025).

The survey’s role in broader black-hole demography is more diffuse but still explicit. In a stacking analysis that combines CEERS, JADES, RUBIES, and GLASS, broad H$\mu$0 is detected in 31% of stacked bins, implying median black-hole masses of $\mu$1 in host galaxies of median stellar mass $\mu$2 (Brooks et al., 24 Nov 2025). That result suggests that the median high-$\mu$3 galaxy host may lie closer to the local $\mu$4–$\mu$5 relation than individually detected extreme AGN, and that individually identified RUBIES-like AGN samples may preferentially trace the upper envelope of the distribution rather than its median (Brooks et al., 24 Nov 2025).

4. Early quiescent galaxies and rapid quenching

RUBIES has become a primary spectroscopic reference set for very early massive quiescent galaxies. At $\mu$6, RUBIES-EGS-QG-1 is spectroscopically confirmed as a compact quiescent galaxy with stellar mass $\mu$7, half-light radius $\mu$8 kpc, and a star-formation history in which most of the mass forms in a short 200 Myr burst followed by rapid, persistent decline in star formation (Graaff et al., 2024). Its deep Balmer absorption, strong 4000 Å break, low current SFR, and AGN-like nebular line ratios make it a benchmark system for early compaction and quenching, and its abundance is reported to exceed current galaxy-formation model expectations by roughly 50–100 in number density at $\mu$9 (Graaff et al., 2024).

At still earlier times, RUBIES-UDS-QG-z7 has emerged as the archetype of the problem. It is described as the earliest massive quiescent galaxy identified to date, at $z$0, only $z$1 Gyr after the Big Bang, with a stellar population inferred to have formed in a single, intense burst beginning at $z$2 and declining rapidly by $z$3 (Turner et al., 19 Sep 2025). In FLARES, only two analogue systems meeting the corresponding observable criteria are found, and their quenching is attributed to AGN that heat and expel gas, suppressing rejuvenation and producing above-average chemical enrichment at fixed stellar mass (Turner et al., 19 Sep 2025).

RUBIES-UDS-QG-z7 is also important because RUBIES-based follow-up has moved the discussion from stellar populations alone to direct gas diagnostics. MgII absorption reveals a neutral outflow with centroid velocity $z$4, an inferred mass outflow rate of $z$5, and a mass loading factor summarized as $z$6, substantially larger than expected from star-formation-driven feedback alone (Valentino et al., 3 Mar 2025). Subsequent ALMA Band 6 observations detect extended [CII] $z$7 emission at the same redshift, with $z$8 kpc, compared to a stellar $z$9 pc, and show that approximately 70% of the [CII] flux arises from a circumgalactic halo (Valentino et al., 19 Jun 2026). Depending on the adopted calibration, the galaxy retains a substantial cold-gas reservoir with $>3$0 and cold-gas fractions $>3$1, despite being approximately 10× more gas-poor than typical star-forming galaxies at fixed redshift and stellar mass (Valentino et al., 19 Jun 2026). The juxtaposition of very low present-day star formation with abundant circumgalactic cold gas has therefore shifted the emphasis from simple gas exhaustion to redistribution, AGN-driven expulsion, and low star-formation efficiency (Valentino et al., 19 Jun 2026).

5. Dusty star-forming galaxies, UV slopes, and the mass–metallicity relation

Although public discussion of RUBIES often centers on LRDs and quiescent galaxies, the survey is deliberately broader. A case study of four submillimeter-selected, ALMA-detected dusty star-forming galaxies at $>3$2 shows how RUBIES NIRSpec resolves heterogeneous evolutionary states inside a photometrically similar DSFG sample (Cooper et al., 2024). All four are infrared-luminous with $>3$3, massive with $>3$4, and heavily obscured with $>3$5 mag, yet one of the four, RUBIES-20125, exhibits a strong Balmer break, H$>3$6 in absorption, and recombination-line SFRs of only $>3$7, in contrast to an IR-inferred SFR of $>3$8, marking it as a post-starburst or transitioning system (Cooper et al., 2024). The same paper derives a nebular attenuation curve that is broadly Calzetti-like in shape but has $>3$9, which it interprets as potentially indicating smaller dust grains or different star–dust geometry (Cooper et al., 2024).

RUBIES also serves as a calibration set for photometric diagnostics that are widely used beyond the survey itself. In a sample of 53 galaxy spectra at $<27$0 drawn from CEERS and RUBIES, NIRSpec/PRISM spectra are used to define “true” UV slopes $<27$1, against which several NIRCam-only estimators are tested (Morales et al., 3 Jul 2025). The main result is that photometric power-law fitting over rest-frame 1500–3000 Å best reproduces spectroscopic $<27$2, with mean offset $<27$3, while single-color methods—especially colors immediately redward of Ly$<27$4—show the largest bias and scatter and are explicitly not recommended when more than one color is available (Morales et al., 3 Jul 2025). In practice, this means RUBIES spectroscopy is being used not only for direct science but also to define best practices for interpreting larger photometric samples.

At lower redshift than the reionization-era programs, RUBIES has also been used to derive a forward-modeled mass–metallicity relation at $<27$5. A sample of 193 galaxies with secure redshifts and measurable R3 and S2 line ratios is analyzed in a fully Bayesian framework that propagates full metallicity posteriors and models both the survey selection function and emission-line observability (Lewis et al., 2 Dec 2025). Two specific biases dominate the result: proper treatment of non-Gaussian metallicity uncertainties lowers the derived MZR normalization by 0.2 dex and flattens the slope by ~20%, while forward-modeling the effect of emission-line observability steepens the slope by ~15% (Lewis et al., 2 Dec 2025). By contrast, the paper finds that the RUBIES selection function itself has a negligible effect on the measured MZR, a useful validation of the survey design for chemical-evolution work (Lewis et al., 2 Dec 2025).

6. Foreground contaminants and survey legacy

RUBIES’ selection for bright, red infrared sources inevitably admits Galactic contaminants, and this has itself become a scientifically productive by-product. A spectroscopic inspection of 3194 public NIRSpec/PRISM spectra in RUBIES identified three brown dwarfs—two likely L dwarfs and one late T dwarf—whose photometry had mimicked high-redshift galaxies or LRDs (Tu et al., 28 Jan 2025). That paper estimates a brown-dwarf contamination rate of approximately 0.1% in a RUBIES-like spectroscopic sample and argues, together with photometric studies, for a true deep-field contamination rate in the range 0.01%–0.1% (Tu et al., 28 Jan 2025). It also notes that cooler T dwarfs overlap the LRD locus, while hotter L dwarfs contaminate the broader high-redshift galaxy cloud (Tu et al., 28 Jan 2025).

A later RUBIES brown-dwarf study expands the sample to seven distant L- and T-type dwarfs with spectral types L1 to T8 and spectrophotometric distances 800–3,000 pc (Morrissey et al., 3 Nov 2025). Three lie more than 1 kpc above the Galactic plane, and two—RUBIES-BD-3 and RUBIES-BD-5—are singled out as thick-disk or halo candidates on the basis of subdwarf-like spectra or metal-poor atmospheric fits (Morrissey et al., 3 Nov 2025). These papers show that RUBIES functions simultaneously as an extragalactic census and as a high-latitude Galactic spectroscopy survey sensitive to rare ultracool dwarfs.

Taken together, the survey’s outputs suggest a particular kind of legacy. RUBIES is not a single-purpose experiment but a spectroscopic platform whose design—bright/red selection, tractable selection function, homogeneous PRISM+G395M coverage, and deep-field JWST context—has made it central to several current debates: the abundance and physical nature of LRDs, the timing and mechanism of early quenching, the interpretation of red $<27$6 galaxies, the calibration of photometric estimators, and the correction of high-redshift chemical-evolution measurements for survey and observability biases (Hviding et al., 5 Jun 2025, Lewis et al., 2 Dec 2025). A plausible implication is that RUBIES’ most durable contribution will be methodological as much as astrophysical: it provides a controlled spectroscopic basis for deciding which red sources are truly massive galaxies, which are AGN-dominated compact systems, which are recently quenched, and which are foreground interlopers.