The color counter model is as follows: model an integer as a collection of positive and negative particles. I will use an ordered pair $(P,N)$ to denote the number of positive and negative chips present.
The interesting thing here is that different pairs can represent the same integer. $(3,0) = (4,1) = (5,2)$, and $(0,2) = (1,3) = (2,4)$ for instance. $(a,b)$ is equivalent to $(c,d)$ if and only if $a+d = b+c$.
This is actually the official definition of the integers when you build them up from scratch. As such it is logically prior to the number line from a rigorous development. It also rhymes with the definition of equivalent fraction: in the foundations of mathematics, we actually define the rational numbers as equivalence classes of pairs of integers (ordinate not being equal to zero) with $(a,b) ~ (c,d)$ if and only if $a*d=c*d$.
This is all far too abstract for school children, but they can learn to "cancel" pos/neg pairs and "create" pos/neg pairs when it suits them. This is valuable work, and can support learning the rules for symbolic manipulation of signed numbers.
I think that both models are valuable. Negative numbers are strange beasts, and it takes a lot of work to understand their internal consistency, how their behavior is the "only sensible" way to make the properties of arithmetic work out, how they can sensibly model real world phenomena (temperature, debt, etc).