From here, the timestamp - when called once a second each second - gets observed as not changing for some time. For instance, it reported being “23:19:36” until “23:20:09” - around 33 seconds. I read this to mean the timestamp likely observes a particular most recent update, rather than the date of request. (Which it should - the date of request or response ought to be an http header rather than in the body of the response.)
Working from http://www.worldtimeserver.com/current_time_in_GB.aspx (Because the Royal Observatory doesn’t publish what time it believes it is) I notice that the the timestamp is consistently upto 40 seconds behind that ‘real’ time. (And that ‘real’ time is within six seconds of the ‘real’ time reported by in.pool.ntp.org.)
My initial feeling would be that your computer’s time is behind the ‘real’ time by about a minute.