RCP Averages.
Trump has been leading in the (relatively meaningless) popular vote averages for weeks, although the amount dropped a while back and looks like it's within margin-of-error. More importantly, he has been consistently leading in 5 - usually 6 - of the 7 battlegrounds, and was even leading in PA (ie. all 7) - which Biden has mostly held only by a very small amount - a short while back. (For those not attuned to US presidential elections: the result is decided by the EVs - electoral votes - of the battleground states, not the popular vote.) He has consequently been leading in the EV electoral map.
"Battle for Congress" shows Senate Republicans at 49 and Democrats at 43, with 8 rated toss-up. Both the Republican and tossup count have nudged up slightly over the past few weeks, which amounts to slipping Democrat fortunes. Because of the way Senate elections are staggered and voting patterns have evolved over the decades, this is the year in the cycle that Democrats are defending a lot of seats compared to Republicans.
Generic ballot (ie. House) is a tie. Republicans have slipped from holding a slight lead over the past few weeks.
A behaviour of US national polls (president, House) customarily remarked upon is that there's a slight Democratic bias (ie. the samples are slightly tilted pro-Democrat; the estimate is almost always a value between 1% and 2%). What that means, for example, is that a poll claiming a "D+1" advantage should be regarded as 50/50. An actual 50/50 result means Democrats are slightly behind.
So, again, without relying on some single result which reinforces what one wishes to believe: Trump looks like he will win; the Senate looks comfortably Republican; the House is toss-up.