I'm assuming you have access to a Matlab license or are a student. Otherwise Matlab is far too expensive to be used for personal projects. The company I work for has several development tools that rely on Matlab. We make Rapid prototyping and calibration tools used in the engineering new vehicles. Our Matlab license costs are substantial.
You really haven't given us enough information to help you. Your choice in processors really has to do more with the algorithms you're trying to run. Matlabs embedded coder can create C code that can run on just about any processor. We've used it to generate code for everything from 8 bit micros up to the latest and greatest 64 bit processors. The question is how complicated is the resultant code and how often do you need it to run? Do you need to support double precision floating point or can you get away with fixed point and integer math?
You will also need to decide what OS you will be running the models on. On some processors You will need to create an application that can wrap around the code generated by Matlab. You will want a fully supported OS driver set. In the Freescale world, the low/no cost choices are Linux, MQX or an OS you develop yourself. We use Linux and an in house developed RTOS we've been porting forward for at least 15 years on the embedded side. We market a windows data acquisition program that runs on our customers laptops and talks to our and our competitors embedded tools though multiple communication protocols.
You need to pick your tools based on the OS you will be running and the board you pick.
If you can get away with using the MQX OS or an in house OS, and don't need a double precision floating point, you can stay in the Kinetis family. If you need Linux and all the benefits it brings, you will want to look at the Vybrid or I.mx based development boards. The Kinetis product can use the free Freescale KDS IDE. The Vybrid and I.mx boards come with support tools for the Linux development environment.
If this is a personal experimentation kind of thing, I would recommend the use of one of the available Linux based boards Matlab supports directly. The Pandaboard, Arduino, or beagleboard. These boards are all ARM core based. The windows version of Matlab can build standalone apps for these boards. You won't need to write the wrapper application. They have prebuilt IO blocks that can access drivers in the OS. I would recommend this as a first step even if you do intend to create your own boards. These development boards are cheap, easy to use and will give you a great deal of experience with the Matlab->embedded board environment. You'll need to check to see if Mathworks or Freescale has evolved any of their support for Freescale development boards to this level.
Norm