Premiere has a match source feature that matches the quality of your source footage so when you upload it to youtube it looks clean and crisp with no quality loss, no hassle having to change bitrates and stuff it sets it up automatically. I don't even need to change stuff it's so easy to export with Premiere. Your mention 3gb files? That is massive a 3gb 60fps 1080p file for me would be like 20+ min of play. I take it your recording in Fraps or DXtory which is your issue. Your life will be more easy with doing a high bitrates with OBS which is more for efficient and resource light compared to any raw recorder. Quality is very comparable on youtube at half the size. Youtube compresses crap out of larger files so having super good quality once compressed is comparable to really good quality video. Most cant tell difference.
Proof:
I'm well aware of Premiere's "match source" feature. However, it doesn't optimize the encoding per YT's standards. It also doesn't match the "quality" of your source, it matches the SETTINGS, meaning the video size, framerate, color depth, etc. Is it a huge difference? No, but it's a difference and I prefer the higher quality output.
Secondly, while it's true that I record in lossless, that has nothing to do with the size of the output file. If you encode a video at 16mbps it doesn't matter how big the input file is, bitrate is bitrate. You could make a video that's a 200kb JPG sitting on screen for 20 minutes and if you encode it as a 16mbps video it'll come out at a few gigs. It is 100% irrelevant what size the input video is when it comes to output size. Period. That's why it's a constant bitrate, that's how rendering works. Try it for yourself. I've seen people on here get confused over this very thing because they had a game video file under a gig but the file Premiere rendered was bigger. Reason? Bitrate.
You're also backwards on "resource light". OBS using compression takes more resources because it's using your CPU to encode and compress the video as it records (GPU if you're using ShadowPlay or something that uses the NVENC codec), whereas a good lossless codec will use effectively no system resources. Granted, even Lagarith and UT still do a LITTLE compression but that's negligible. Sure, the finished video is a lot smaller for you, but... so what? We're both rendering to the same size. My 250gB gaming video or your 2gB one are going to come out to the same size if Premiere encodes it at 10mbps. Not to mention OBS is resource heavy to begin with. Not quite as bad as XSplit but it's up there.
Fourthly, DXTory doesn't make "big files" while OBS makes "small" files. You can get a giant file with OBS or a tiny one with DXTory, what's important is the codec you use. I indeed go lossless with UT Video because it means minimal compression along the way. YouTube compresses the file, sure, but why would I compress the crap out of it myself if I can avoid it? Lossless video, high bitrate per YT's standards, and that gives me the highest quality video possible at the end so when YouTube does their magic, I barely lose anything.
What's the downside? An extra five minutes of upload time? What exactly is the problem? Now keep in mind, I am NOT telling you that you have to do things the way I am doing them, but I will say that giving YouTube the highest quality file and putting the best file possible into your editor is always optimal as long as you have the available system space. If you don't, then pick the best compression tool which, in my experience, is NVENC if you have an nVidia card.
But remember, whether you do lossless, Bandi, x264 or NVENC, that has nothing to do with the final size of your video unless you fiddle with the bitrate.