Actually, you only need IR. I have been a certified thermographer for many years and done literally thousands of measurements in hundreds of applications. You can make very accurate and sophisticated measurements in complex situations using IR if you know what you are doing. It's more than just pointing at the surface and taking a reading.Thanx for that. If one were to have a really good thermometer, it would be necessary for it to measure IR AND visible. As when a molten metal is white hot, you can SEE the white, and that is not IR--there is plenty of IR there, but it will not be measuring the best form of light to use which will be in this case, visible. IR would be best when the metal is in a dull red to cherry red.
Most of the industrial IR pyrometers and cameras use a mid-wave sensitive detector (7.5 to 13 micron). This is adequate for most industrial and commercial applications. Special applications may need a different wavelength. For example, if you need to measure through glass, such as a filament in a light bulb, mid-wave won't work as glass is an IR reflector at those wavelengths - you'd need to use a detector that measures in the 2-5 micron range.
Calibration is also important - a properly calibrated mid-wave IR camera, for example, can make accurate measurements up to 1200C. This type of calibration is done using a series of black body sources - not something the manufacturer of a typical low end handheld pyrometer is going to do.
Back to the OP's original question: no, a cheap IR pyrometer isn't going to give you an accurate reading. Could it be done with IR or other non-contact method? - Maybe, but it would be expensive and complex.
I do some metal casting, but the melt volumes are small - I use an electro melt furnace with a PID controller, so I know exactly what the temperature is.