When comparing against something like a Pixel running GrapheneOS, it's honestly a bit more puzzling to me. Granted, I'm definitely not the audience for this, but with G_OS you can do most things that a regular phone can do, without taking several minutes to install Firefox.
As much as I love privacy (going as far as having a semi-random username), this phone is a bit puzzling. I hope someone can throw more light on this.
I'm not familiar with GrapheneOS but I assume it follows the usual model when repurposing Android devices of taking various closed source blobs (i.e. drivers etc) and rebuilding the open source bits around them? If so, this approach usually locks you into a Linux kernel version to remain compatible with the blobs which limits you on kernel features and fixes as well as who knows what exposure the blobs have to offer, which also will likely never get updates.
Sticking to an LTS kernel branch for the lifetime of the device isn't due to anything closed source. GrapheneOS only supports devices with proper security support for all the firmware, drivers, etc. and again there are no closed source kernel drivers. We can support pretty much any mobile device with alternate OS support since any serious one will have AOSP support. Most devices have lackluster security and don't meet our requirements. We're working with a hardware vendor to get a non-Pixel phone actually meeting reasonable security requirements.
Librem 5 has a bunch of components where they are not shipping updates. You have things very much backwards on that front. The Librem 5 does not come close to meeting the security requirements to run GrapheneOS. It has a bunch of poorly secured and insecurely configured legacy hardware often without proper updates available, components that are not properly isolated via IOMMU, no secure element or all the stuff that comes along with that (HSM keystore with a nice API used by apps, Weaver to make disk encryption work for users without a high entropy passphrase like 7 diceware words, insider attack resistance, working attestation not depending on hard-wiring hashes and a lot more) and many other things. The OS they use has a near total lack of any systemic overall privacy/security work or privacy/security model and only falls further and further behind. The most exciting feature for securing devices right now is hardware memory tagging support in ARMv9, but there are years and years of tons of important privacy/security work done in a systemic way across hardware/firmware/software which are missing there before worrying about stuff like that.
Marketing something as private/secure and spreading tons of misinformation and outright lies about the mainstream options doesn't make it secure or more secure than those. It's actually pretty funny that they mislead people about the isolation of hardware components like the cellular baseband in other devices when the vast majority of mainstream phones (iPhone, Pixel, Qualcomm SoC devices, Exynos SoC devices) have it done quite well when they don't. Strange that they get away with these games of misrepresenting things, hiding the fact that they still have entirely proprietary hardware and near entirely proprietary firmware for the SoC and other hardware components, etc. Hiding proprietary stuff doesn't make it go away. Not updating it doesn't make it go away and simply ensures a highly insecure device.
Purism's marketing material is outright deceptive, e.g. they insist that in competing phones the baseband blob has access to system memory, which is a lie. The reality is that the baseband blob in the Librem 5 (which is every bit a giant blob as that in the competition) has access to the USB port of the AP and there is no filtering implemented yet, so the attack surface it is exposed to is every USB driver in the Linux kernel, which is much worse than systems with embedded basebands and proper memory firewalling where the baseband has no more inherent access, but is exposed to a smaller attack surface. That means that you are more vulnerable to giant blobs doing evil things with a Librem 5 than with, say, an Android phone running a free OS build.
Then there's the whole hilarious situation with the RAM initialization blob where Purism went and hid it behind two layers of CPUs (not execution, just handling it), because somehow doing that - which provides absolutely no benefit to the user, it's just a waste of engineering time - made it possible to certify the phone under the FSF's utterly broken and nonsensical "Respects your Freedom" program, even though precisely zero freedom was gained by doing this, since it still running the same blob on the same final CPU with the same access. All the while while reducing security, since the blob is then made no longer part of the normal OS image and is not validated with it, so it could be backdoored as part of a supply chain attack and you would be none the wiser.
The whole thing just stinks the more you look into it, and it is completely evident that the folks behind Purism are a mix of deliberately deceiving people and just clueless about security and modern embedded platforms.
You're saying that as if the firewall handled the communication with the modem. There's a Linux driver behind the firewall to do the actual communication, except that's probably not a USB driver.
The attack area is probably comparable, except a misconfigured USB driver doesn't automatically give full memory access, while a misconfigured IOMMU (the firewall) does.