The highest casualty rate I’ve ever seen published for nuclear war was somewhere around 40 to 50% of the population of the US. Interestingly, despite a nuclear strike of over 25,000 nuclear weapons, Russia was expected to win that one with less than 25% of their population killed.
And there is no proven scientific basis for a nuclear winter to be the results of nuclear war. Even less so today, considering that the United States and Russia have far far fewer nuclear weapons then they did in the past. Russia only has a few thousand functional nuclear weapons, most of which are not in a state that could actually be deployed in a war.