The sun subtends an angle of about 0.5 degrees, and a pixel on Lepton 2.5 has an iFOV of 0.6 degrees, so the sun cannot completely fill a pixel. Most likely the image would only be partially filling two or more pixels, so not presenting the full radiation of the sun on any one pixel, although it would probably still saturate these pixels. Assuming that Radiometry is on, the camera is in high resolution T-linear, and is in high gain mode, 45,000 counts would be 177 degrees C, which would probably be saturation for those pixels unless the camera is very cold. The actual sun is hotter, of course.
A value of 23,000 assuming the same configuration would indicate -43 degrees C, which is also about where the camera would be expected to saturate, unless the camera is very hot. The actual deep sky would be colder.