you could possibly do it only with an accurate temperature probe if you are using heating elements and are recirculating. you can accurately measure how many watts of power are going into the water, and you can calculate the rate of heat loss and thermal mass of the container.

take two temperature readings spaced x-amount of time apart. subtract the amount of heat lost during that period of time and factor in the thermal mass of the container. the amount of temperature change over time, plus the amount of energy input, will let you calculate how much water is in the container.

similar theory to volumetric measurements using the pressure readings of a gas. there are probably better ways to do it, and there are obvious downsides that can easily throw the accuracy off, but if those are accounted for then all you need are temperature readings, power consumption measurements and some math.