A standard Shakura & Sunyaev (1973) thin accretion disk radiates as a local blackbody at every radius, with an effective temperature profile that falls off steeply with distance from the inner edge:
T(r) = T_in · (r_in/r)^0.75 · [1 - √(r_in/r)]^0.25
The r-0.75 power law is the reason a real accretion disk shows a sharp color gradient: the innermost annuli — sheared fastest by differential (Keplerian) rotation — dissipate far more heat per unit area than the slow-moving outer disk, so the disk runs from blinding blue-white/UV near the inner edge to dull deep red at the rim. The bracketed term forces T → 0 exactly at the inner edge itself (the torque-free boundary condition), which is why the very brightest ring sits just outside r_in rather than on top of it.
The scale temperature follows from energy conservation per unit area, σT*⁴ = 3GMṀ / (8π r_in³) with r_in = 6GM/c² (the innermost stable orbit of a non-spinning black hole). If the accretion rate is a fixed fraction ṁ of the Eddington limit, then Ṁ ∝ ṁ·M and T* ∝ (ṁ / M)^0.25: heavier black holes have cooler discs (X-ray for stellar-mass holes, ultraviolet for supermassive ones), and a higher ṁ makes the disc hotter. The hottest ring sits at r = 49/36 r_in and has Tmax = 0.488 T*.
Honest simplifications. Newtonian potential with a Keplerian disc and a torque-free inner edge; the radiative efficiency is the Newtonian value 1/12 (the full general-relativistic value for a non-spinning hole is about 0.057). No relativistic beaming, gravitational lensing, colour correction or corona. Each ring is drawn with an approximate blackbody colour, not a true spectrum; every disc hotter than about 10 000 K looks blue-white in true colour, so the false-colour button maps log T onto a thermal palette instead. The disc is cut at r_out = 10 r_in, so it radiates only about 76 % of the infinite-disc luminosity; the readout shows exactly that. The star field and the black-hole sprite are Blender 5.2 renders used for decoration; the sprite is drawn at the true horizon size (2GM/c² against r_in = 6GM/c²).
Check. The luminosity inside r_out is integrated numerically (Simpson, 4000 steps, 2 faces, σT⁴ at every radius) and compared with the closed form L = (3GMṀ / 2r_in) [(1 − 1/R) − (2/3)(1 − R^-1.5)], R = r_out/r_in; for the infinite disc it equals GMṀ/(2r_in).