I'm planning on putting a FAQ together, here's the section I wrote on version differences:
The users of the I-Appliance message boards have adopted a version scheme that IS NOT used by Netpliance. It is a version system used by the hackers to distinguish changes made by Netpliance to the hardware/software of the I-Opener. The original I-Opener is referred to as a v1 (Version 1) and contains the sought-after BIOS, which will boot any operating system. The earliest v1 I-Openers shipped with a 180MHz Winchip, overclocked to 200MHz. The Winchip used was single voltage (3.3v) and clocked at 3.0x 66MHz. The v1 also had many security holes in its installation of QNX, which allowed easy access to a command prompt. Shortly after Netpliance started modifying the I-Opener, v1 I-Openers with clipped IDE pins started surfacing - they were not given a unique version number and are thus far are very rare.
V2 I-Openers were the first to arrive to hackers who had placed orders early from Netpliance and to those lucky enough to catch the first wave of I-Openers arriving at CompUSA. These v2 units contained the newly-modified BIOS which was annoyingly glued down to avoid removal. Oddly enough, the v2 contained the same insecure version of QNX that the v1 shipped with. A service called BadFlash started offering replacement and reflashing of I-Opener BIOS chips and many a v2 hacker "degooped" their BIOS chip and sent it away or replaced it. It wasn't long though before the insecurity in the v2's QNX was exploited by a root password decryption program and later a QNX-based BIOS reflashing program called QNXFLASH written by Smogbound. These exploits allowed the BIOS in the v2 to be "downgraded" back to the v1 BIOS WITHOUT the need to remove the chip. Obviously, Netpliance did not like this and decided to close up those holes in their distribution of QNX - and so the v3 was born.
The v3 contained the same BIOS version and epoxy as the v2, but now closed the exploits used to obtain a command prompt. The v3 could be identified by its non-interactive demo with a female narrator, whereas the prior demos were interactive and narrated by a male. Shortly thereafter, the v3 was defeated by connecting a hard drive containing the v1/v2's version of QNX to the IDE header and allowing it to boot - this made reflashing using QNXFLASH possible again. Netpliance again responded to this threat by releasing what became known as a "v3b" - a v3 with security Torx screws. The hackers responded by drilling out the torx pin and using a T8 screwdriver.
The next version of the I-Opener, the v4 was not borne of anti-hack modifications, it came to be because the supply of Winchips and Sharp DSTN screens had dried up. This new version contained a DSTN screen manufactured by Hitachi and a PR266 RISE (not to be confused with RISC, the RISE is a x86 compatible) dual voltage (3.3vIO, 2.8vCore) CPU clocked at 2x 100MHz. The v4 also was being shipped with no epoxy on the BIOS, and even though many rumors were started - it worked just fine when replaced with the original v1 BIOS. While the epoxy was now gone, the Torx screws remained. An issue with the v4 did arise when using the "v1/v2 QNX on Hard Drive" method of reflashing the BIOS - qnxflash would not work with the RISE CPU. A temporary fix was found by installing a Winchip from a prior I-Opener version for the reflashing process. Later, Smogbound released an updated version (1.02) of QNXFLASH.
Netpliance was not sitting idly by; however, because the present v5 I-Opener now contains a BIOS that will not respond to any known key combination during bootup. It also contains an updated version of the installed QNX as well; presumably to provide something new for the BIOS to identify so it cannot be tricked into booting a prior version of QNX. It has been confirmed by BadFlash that the he new v5 I-Openers WILL function with the v1 BIOS; however, they will no longer boot their installed QNX once the BIOS "downgrade" has taken place. V5 I-Openers also have been said to no longer have a Yamaha audio chip and instead have an Analog Devices AD1881A. This may make obtaining sound in your favorite alternate OS more "interesting".