TOI-6894 (TIC-67512645) was observed by TESS13 during both the primary and extended missions. In the primary mission, TOI-6894 was observed in sector 22 (18 February to 18 March 2020), and in the extended mission, TOI-6894 was observed in sectors 45, 46 and 49 (6 November to 30 December 2021 and 26 February to 26 March 2022). Across all sectors, TOI-6894 was observed in the FFIs, and so TESS photometry is available at a cadence of 30 min for sector 22 and 10 min for the extended mission sectors. The TESS FFI photometry was processed by the TESS Science Processing Operation Center (SPOC)46. We accessed the data through the TESS-SPOC High-Level Science Product47. For our analysis, we used the PDCSAP light curves, which have been processed to remove spacecraft-related instrumental systematics48,49,50. The TESS light curves for TOI-6894 are displayed in Fig. 1. We display a cut-out pixel image of the area surrounding TOI-6894 in Supplementary Fig. 1.

TESS candidate detection

TOI-6894 was included in a systematic transit search for giant planets with low-mass host stars in the FFI data from the TESS primary mission14. In short, this search detected periodic transit-like signals using the Astropy implementation of the box-fitting least squares algorithm51,52. It excluded clear false-positive scenarios and performed a transit-fitting analysis to identify probable giant planet candidates. Following these automated steps and some further manual vetting, TOI-6894 b was identified as a good quality giant planet candidate14. TESS-SPOC independently identified the signature of TOI-6894 b in transit searches of the FFI data from sectors 45, 46 and 49 using an adaptive matched filter53,54,55. After vetting the results of sector 49 with a modified version of TESS-ExoClass (https://github.com/christopherburke/TESS-ExoClass) for FFI targets47, TOI-6894 b was reported as a candidate15. The difference image centroid analysis56 for sector 46 constrained the location of the target star to be within 4.3 ± 2.5 arcsec of the transit source, substantially reducing the possibility of a nearby blended eclipsing binary scenario. TOI-6894 b was made a TESS object of interest on 1 February 2024.

ExTrA observations

A full transit of TOI-6894 b was observed by ExTrA57, a low-resolution near-infrared (0.85–1.55 μm) multi-object spectrograph, on 25 April 2023. ExTrA was fed by three 60-cm-diameter telescopes at the European Southern Observatory’s (ESO’s) La Silla Observatory in Chile. Five fibres were positioned in the focal plane of each telescope to select light from the target and four comparison stars. Owing to the faintness of the target (J = 13.2 mag), we used the low-resolution mode of the spectrograph (R ≈ 20) and employed fibres with a 4″ aperture to minimize the contribution of sky emission. The resulting ExTrA data were analysed using custom data-reduction software. The transit light curves from the three ExTrA telescopes are presented in Fig. 1 and Extended Data Fig. 1.

SPECULOOS observations

Six full transits of TOI-6894 b were observed using various telescopes in the SPECULOOS58,59,60 1m0-network at ESO Paranal Observatory in Chile and Teide Observatory in Tenerife61. All telescopes were equipped with a deep-depletion Andor iKon-L 2k × 2k CCD camera with a pixel scale of 0.35″, resulting in a total field of view of 12″ × 12″. We collected the data during transits of TOI-6894 b on the nights of 2, 12 and 19 February 2024 in the I + z’, Sloan-g’, Sloan-r’ and Sloan-z’ filters and during an occultation of TOI-6894 b on the night of 7 February 2024 in the Sloan-z’ filter. Science image processing and photometric extraction were performed using the PROSE pipeline62 (https://github.com/lgrcia/prose). The SPECULOOS data were detrended using external systematics variations related to time, the full-width at half-maximum of the point spread function, the sky background, the airmass and the X and Y pixel positions. The entire SPECULOOS transit photometry is plotted in Extended Data Fig. 1, and a selection is plotted in Fig. 1. The SPECULOOS occultation observation is plotted in Extended Data Fig. 4.

TRAPPIST observations

A full transit of TOI-6894 b was observed with the TRAPPIST-South63,64 telescope on 12 February 2024 in the blue-blocking filter with an exposure time of 140 s. This is a 60-cm robotic Ritchey–Chretien telescope installed at ESO’s La Silla Observatory in Chile. It is equipped with a thermoelectrically cooled 2k × 2k FLI Proline CCD camera with a pixel scale of 0.65″ and a field of view of 22′ × 22′ (refs. 63,64). Science image processing and photometric measurements were performed using the PROSE pipeline. The TRAPPIST photometry is plotted in Extended Data Fig. 1.

Sierra Nevada Observatory observations

We observed TOI-6894 b on 19 February 2024 using the T150 at the Sierra Nevada Observatory (Observatorio de Sierra Nevada or OSN) in Granada, Spain. The T150 is a 150-cm Ritchey-Chrétien telescope equipped with a thermoelectrically cooled 2k × 2k Andor iKon-L BEX2DD CCD camera with a field of view of 7.9′ × 7.9′ and pixel scale of 0.232″. We used the Johnson–Cousin I and V filters simultaneously with exposure times of 120 and 90 s, respectively. The photometric data were extracted using the AstroImageJ package65 and are plotted in Extended Data Fig. 1.

LCOGT observations

TOI-6894 was also observed from the South African and Tenerife (Teide) nodes of the Las Cumbres Observatory Global Telescope network (LCOGT)66 using the 1-m telescopes on 19 February 2024. Both observations were carried out alternately in the V and z s bands with exposure times of 300 and 70 s to cover the full transits. The observations were done with Sinistro cameras, which have a field of view of 26′ × 26′ and a pixel scale of 0.389″. The raw images were automatically calibrated using the BANZAI pipeline67. We then performed the photometric analysis using the AstroImageJ software65 with an 8-pixel (3.1″) or 5-pixel (1.9″) aperture. The estimated point spread functions of the two observations are 1.85″ and 1.65″, respectively. All the LCO photometry is plotted in Extended Data Fig. 1, and the V-band photometry is also plotted in Fig. 1.

MuSCAT2 observations

A full-transit observation of TOI-6894 b was collected on 19 February 2024 ut using MuSCAT2 (ref. 68) mounted on the 1.52-m Telescopio Carlos Sánchez at Teide Observatory, Tenerife, Spain. MuSCAT2 is a multicolour imager with a field of view of 7.4′ × 7.4′ and a pixel scale of 0.44″. The observation was carried out simultaneously in four bands (g, r, i and z s ). However, the g- and r-band data have a low S/N due to the large scatter induced by clouds. Therefore, we excluded these two datasets in our analysis. The i- and z s -band data were also impacted by the clouds but still had a sufficient S/N to be usefully included in the analysis. We carried out aperture photometry using the MuSCAT2 pipeline69 after dark-frame and flat-field calibration. The pipeline automatically finds the optimized aperture to minimize the photometric dispersion and then fits a transit model after accounting for instrumental systematic effects. The MuSCAT2 data are plotted in Extended Data Fig. 1.

FIRE and Magellan

TOI-6894 was observed on the night of 26 February 2024 with the FIRE16 intermediate-resolution spectrograph operated at the 6.5-m Magellan Baade telescope, Las Campanas Observatory, Chile. We used a 0.6 × 7 arcsec slit that provided a spectral resolving power R = 4,500 in the wavelength range 0.82 < λ 5 R ⊕ ) in short orbital periods (P < 6 days) with recovery rates of ~100%. Indeed, TOI-6894 b falls in this region. These planets become more challenging to detect for longer orbital periods, although doing so is still possible, with recovery rates between 40% and 80%. These results allowed us to conclude that the existence of such planets in the system is very unlikely. On the other hand, small transiting planets with sizes smaller than 4 R ⊕ would be undetectable in the complete set of periods explored. Hence, we cannot offer any constraint on the existence of these planets in the system.

Atmospheric characterization prospects

We expect TOI-6894 b to become a benchmark planet in the study of temperate H/He atmospheres. TOI-6894 b receives a stellar irradiation S = 5.50 ± 0.44 S ⊕ , which translates into an equilibrium temperature T eq = 417.9 ± 8.6 K. This value assumes an albedo A = 0.1, like that of many hot and warm Jupiters115. At this temperature, it is widely expected the planet is dominated by methane chemistry, like WASP-80 b (refs. 43,44). Using these properties, we modelled possible atmospheres with and without clouds and with high and low C/O ratios, and we found that methane absorption features in the transmission spectrum of the planet would be expected to have amplitudes of 6,000, 9,000 and 11,000 ppm in the optical, near-infrared and mid-infrared, well in excess of any other giant to date, particularly for planets with a similarly low equilibrium temperature. This is mainly caused by two effects. The host star, TOI-6894, is small and transmission features are amplified by R * −2 and by the surprising low surface gravity of TOI-6894 b. To address the detectability of individual molecules within the TOI-6894 b planet spectrum, we used the methodology applied in refs. 116,117,118 for the transit geometry. We ran the James Webb Space Telescope PandExo noise model across a grid in number of transits from 1 to 100, which is sufficient to establish a simple S/N scaling relation, and we determined the S/N on the difference between the model spectrum and the fiducial spectrum. Our PandExo simulations of observations using the NIRISS/SOSS, NIRSpec/G395M and MIRI/LRS modes on TOI-6894 b showed that a single transit could suffice to retrieve abundances of key atmospheric species like methane, water and carbon dioxide, with a total expected S/N ≥ 100. We plot example transmission spectra obtained from PandExo in Extended Data Fig. 8. Furthermore, as illustrated in Extended Data Fig. 8, molecular absorption features should be detectable at wavelengths beyond 2 μm, even with a cloud deck at 1 mbar.

To place this planet in context, we calculated its TSM (see ref. 45 for details). The TSM can be used as a measure of how amenable a planet is to atmospheric characterization through transmission spectroscopy. We found that TOI-6894 b has a TSM of 356 ± 58. Comparing the TSM of TOI-6894 b to values for other known planets (Fig. 3 and Extended Data Fig. 7), we found TOI-6894 b to have the highest TSM of any giant planet with a host star less massive than 0.7 M ⊙ and the second highest for any planet with a low-mass host star (M * ≤ 0.4 M ⊙ ), second only to GJ 1214 b. TOI-6894 b particularly stands out when considering other planets with a low equilibrium temperature.

Assuming that the planet has an albedo A = 0.1, we would expect its emission spectrum to be highly amenable to the detection of atmospheric features. As for transmission, we modelled possible emission spectra and found typical eclipse depths of 1,000–6,000 ppm in the mid-infrared. Studying the atmosphere of TOI-6894 b could provide easy access to an H/He atmosphere intermediate between those of hot Jupiters and the Jupiter in our Solar System. Studying its chemistry may help to refine atmospheric models. In addition, studying the atmosphere of a planet can provide further clues related to its formation history. The star is metal-rich ([Fe/H] = 0.142 ± 0.087), and it will be of interest to measure whether its atmosphere is too. Such measurements would reveal the true metal content of TOI-6894 b, thereby also revealing the composition of TOI-6894 b and giving clues about its formation history38.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.