... wrote ...

For higher powers I would recommend putting a thermistor on the tec, and running it in a a control loop (for simple experiments a simple proportional control would probably work fine but a full pid loop would speed up the response considerably). You can compare the power it takes to keep the tec at a constant temperature with no load vs what it takes to keep it at the same temperature with the laser shining on it, which you could then use to extract the laser power using the specifications from the tec data sheet. Although due to various nonlinearities you really are better off dumping the tec altogether and just using a resistor mounted to the heatsink, and keeping the heatsink at a constant temperature. By looking at the difference in power fed to the resistor with the laser on and off you can accurately extract the laser power.