When I called the MSI Claw an embarrassment in my May 18th review, one commenter pointed out they’d gotten higher benchmark scores than me. Sure enough: MSI had pushed out new graphics drivers and BIOS updates between the time I benchmarked and the time we published.

I’ve now tested those updates. In some of my tests, they’re better. In others, I actually found them worse! Overall, they don’t change my conclusion one bit: the MSI Claw is the worst pick from today’s crop of consumer handheld gaming PCs, and you should steer clear.

Below, you’ll find my new benchmark results in the same games, across all of the Claw’s standard power modes. In almost every test, you’ll see the Claw dramatically behind the competition in speed while using more power.

Worse, I’m still finding the Claw varies wildly between playable and completely unplayable in demanding games where the less expensive Steam Deck has no issues. In Shadow of the Tomb Raider, I saw the game repeatedly and inexplicably dip to just 20fps in a cave with no enemies around at 720p and Medium spec, before recovering to a stable 60fps a little while later.

I saw something similar in Cyberpunk 2077, even though the MSI Claw now generally has a higher framerate in that game after updates.

And I’m also now seeing some very weird behavior with the Claw when it comes to power modes. You might expect MSI’s “Performance” mode to be speedier than its “Balanced” mode, right? That’s not what I saw in Cyberpunk 2077 — and when I took a close look at the framerate dips, I noticed they always coincided with an unexpected power dip.

While the Intel Meteor Lake chip consumed a pretty stable 30 watts in Balanced mode, it tried to hit 35, 36, even 37 watts in Performance mode — and couldn’t maintain it, dipping down to 29 watts again and again. Plugged into USB-C power, the Performance mode worked better, continually pulling between 37 and 45 watts of power. Not that the extra power made an improvement in Cyberpunk, unfortunately.

And if you’re seeing higher scores than I am, I highly suggest you run that benchmark another three or four times, because it looks like MSI now runs its chip unsustainably faster for a while after you switch power modes, then throttles back down. The Asus ROG Ally does the same trick, and I always had to run my benchmarks a bunch of extra times there before I start counting, so I don’t throw off the average with outliers. I didn’t need to do that when I first reviewed the Claw.

To MSI and Intel’s credit, at least Dave the Diver feels playable now at a smooth 60fps with only occasional drops, unlike in my original review.

Overall, I think the MSI Claw should never have gone on sale. I’m glad I never had the urge to buy one when it first came out. I would feel cheated, just like the users in MSI’s own forums say they feel cheated, now that MSI has announced not one, but two improved versions less than three months after the original came out.

It certainly doesn’t help that the company put it on sale before giving reviewers a chance to warn anyone.

If MSI is serious about the handheld space, and it sounds like the company is, I think it should offer a trade-in program for buyers of the original Claw. If you want to fix this, MSI, let them pick one of this fall’s revisions at a massive discount.