Depends on which 8-bit, Sega Master System or NES?

You can't do those sprites on the NES, as sprites can only have 3 colours (2-bit sprites).

But you can do those sprites on the Sega Master System, as it can do 15 colour sprites (4-bit sprites).

So yes those look like sprites on the 8-bit Sega Master System, a console you may not be aware of.

For computers, you can use almost all bit types which is another advantage of computers over consoles.

People are coding for PCs with 32-bit color depth just for the fun of it. I've even heard of 64-bit experiments. Granted, they are not in the mainstream, but you won't see that with consoles at all to the best of my [limited] knowledge. Software solutions for hardware limitations have been used for many years.

I'd classify both as 8-bit. The jump up to 16-bit was a fundamental leap. With 8-bit color, games almost by necessity had to work with a predefined palette of colors. This tended to reduce shading ability and machine-generated effects. Colors tended to also have a greater amount of contrast, although a careful choice of palette could go a long way toward minimizing this constraint. (The original Starcraft was incredible in this regard. It technically had a palettized 8-bit color depth, but stylistically felt like 16-bit to me.)

16-bit color allowed games to treat colors as a composite of red, green, and blue channels, generally with red and blue each getting 5 bits (32 distinct levels), and green getting 6 bits (64 distinct levels) since the human eye is more sensitive to green. This allowed developers and artists to escape from the limitations of predefined palettes, enabling a variety of new options, such as applying machine generated effects, both at production time and game runtime. This also reduced the need to work with or around high contrast colors, enabling a large variety of stylistic options.

I would also avoid classifying either of the two images as 4-bit, even though both could technically be represented with a 4-bit palette. The reason is that outside of clever tricks, the 4-bit limitation applies to the entire scene, not individual sprites. When you only have 16 colors to work with for everything, each individual sprite is unlikely to make effective use of most of those colors. Most sprites would still only use 2 to 5 colors each. (But with care, similar to 8-bit palettes, you could construct a pretty effective 4-bit palette. Warlords II I feel did a great job of this. It had black, white, and 4 shades of gray, 3 shades of green, 2 shades of brown and blue, and just one shade each for yellow, orange, and red. This gave just enough gradient flexibility for the important stuff, while still keeping enough variety in the overall color selection.)

I'd put them as late-gen 8-bit sprites, remember the Master System guys.

128 bit pixels? You're saying that they have 32 bits for each channel plus 32 bits alpha?

32 bit color depth means = 8 bit per channel (RGB) + 8 bit alpha. 32 bit per pixel.

128 bit per color, according to my maths, means 30Mb per frame on a 1080p display. Which means that hardware like that could push 4k resolution (32 bit per pixel) without breaking a sweat and I don't think it works like that just now

And yes, 32 bpp is what consoles and PCs have been using for years now. Both doing HDR in the GPU.

EDIT: Ohhh, google fu tells me this: Each color is stored in a float (ie, 32 bit) for HDR rendering. Though I do have heard of the FP16 format used for HDR. So that means, 8 bit per channel + 8 bit alpha for output, possibly 32 bit per channel for computation.