Nice usage for lambdas is optional callable parameter to a function. You've got a dictionary full of stats while your program runs. Some stats, as new numbers are recorded, you'd like to keep the biggest one:
self.add_stats(key, value, lambda x, y: max(x, y))
Or, you want to simply add the new numbers to existing stats:
self.add_stats(key, value, lambda x, y: x + y)
And in def add_stats(), you'll have code that handles that callable if it exists. X being the existing stat, and Y being the new value being passed in. Sure, you can do this with an array of flags, but lambdas allow for much great flexibility with much less code. Downside is that lambdas are somewhat expensive since you're creating a new function every time add_stats() is called. But, hey, it's python, not exactly known for performance.