https://www.medrxiv.org/content/10.1101/2020.09.16.20194787v1.full.pdf
66%
Manaus is a city of 1.8 - 2 million (on par with Philadelphia, Phoenix or Houston) in Brazil. This paper suggests that they've essentially reached herd immunity, with an overall IFR in the range of 0.25% due to their relatively young population.
There is some drop-off in seroprevalence over time, Manaus may be interesting to watch to see if this leads to reduced immunity, or if immunity holds despite the lower antibody levels. (My money is on the latter, but it's far from clear.)
Keep in mind that the herd immunity threshold varies with population density and other characteristics, including seasonality. Take this 66% as one data point with many more needed.
Tokyo - 46%
https://www.medrxiv.org/content/10.1101/2020.09.21.20198796v1
It's office workers, so likely marginally higher than the general population. The abstract doesn't say if they corrected for the test's false negative rate, which can be significant (like 20%) in some sero tests; I'll have to check the details in the paper.
With serological positives usually not showing up until 2-3 weeks after symptom onset, it's strange that their results peaked a month before the confirmed case peak in Tokyo. (Possibly some spread within the company in the earlier phases of Tokyo's curve?)
Should be some good reading.