No, 128-bit is closer to the truth. According to the OpenGL documentation in MSDN, "current color values are stored in floating-point format, with unspecified mantissa and exponent sizes." This will usually be a float which is usually 32 bits. OpenGL keeps track of 4 color components, so that means 128 total bits per color. But, again, the floating-point format is undefined by OpenGL, nor is it exposed. Most frame buffers these days are 24-bit or 32-bit, but this won''t be true for much longer, since 24/32-bit still produce noticeable banding artifacts; hopefully, frame buffers will be floating-point with at least 16 bits per channel.

just repeating what was in the opengl 1.2 source code from sgi