First and foremost, I am seeing a lot more effort and time being invested into the actual constructions, and a lot less guesswork as far as where things should go. There are however a number of issues that come to mind here. I put some redline notes directly on your work, but I'll list the issues I noticed below as well:

• So I'm noticing that you're drawing your cars in entirely different orientations than the reference images you've shared. Since I mentioned having multiple reference images to help flesh out your understanding of a given vehicle, perhaps you only included one of many for each car, but it definitely raises some questions about whether you're trying to invent an orientation before you've really established a solid grasp of what is actually present in the reference images you've got. If I were to invent my own orientation of something new to me, I'd first start by doing a bunch of studies from actual photos to flesh out my understanding of its nuances and construction.

• In the first car, you appear not to have pinned down the length of your construction with any specificity. I'm not seeing you establishing a unit cube and repeating it back in space (as shown here). You did this correctly in the second construction, so I'm unsure of why you skipped this incredibly important step knowing that in your last set, proportions were one of your biggest struggles.

• While I'm not seeing any orthographic proportional studies here, I certainly hope you've been doing them.

• There are a lot of signs that suggest that you're drawing a lot more from memory - perhaps getting really caught up in your construction, and as a result not looking at your reference frequently enough. This manifests as major elements of the cars that simply don't match your reference material. Taking that further, you're also stopping your construction pretty early - I'm not seeing any of the headlights, grills, etc. which are all elements that can still be constructed, and the base constructions upon which those are meant to be built are really deviating from your reference material, both in proportion, and in general structure.

It's a step in the right direction, but you need to invest more into the direct observation of your source material in order to ensure an amount of accuracy. I don't expect students to create perfect representations of their work, but while your actual constructions are really quite solidly built here (which is excellent), they are fundamentally different vehicles from what you set out to produce.