Some kinds of work are easier to do than others. At one point, I decided that 60 hr/wk of active work seemed the approximate sustainable limit for some typical mix of both hard&easy work, over a period of weeks or months. But even if you can schedule the hard&easy parts for fatigue, 60 hr/wk still has extra costs (e.g., stresses to health, outside relationships, outside activities, internal morale as people get irritable).
I've also done my share of all-nighters and 80-hour weeks, and it's rarely/never worth every hour. I've learned to manage projects and tasks better.
I also have some sense of how performance degrades severely by the 30 hours awake mark, and a sense that there's a ramp up to that impairment before it becomes obvious.
Most recently, I advised a developer to try for 40 productive hours of work per week, and to emphasize being sharp and non-fatigued for any task that would benefit from that. Almost all of their work required them to be sharp. Keep the 60-hour heroics in reserve for emergencies only. (Or for occasional easy flow work, like sketching out a UI skeleton in some new framework. But don't work all night on something easy, and then throw off your sleep cycle for a week when you need to be sharp for harder stuff.)
(The last time I did a weekend work marathon, which required some meticulous reverse-engineering-ish work that I think was already the work-smarter way and couldn't be sped up further, I imagined the focus and pace of an astronaut who was calm and all-business about solving some critical problem in time. That was both inspirational and also set a simple internal watchdog on my mood and how progress was going. Had I been burning stressful 60-hour weeks for months before, I might've been too fatigued/stressed to pull it off without error or poor decisions. Had I not had much experience with working long hours, I might not have appreciated how much discipline they require, and the costs.)