The trademark's already been found.
In any case, the only reliable information we have is that it's a graphing calculator, that it's called Prime, and that its product number is probably NW280AA. And that's not even exactly reliable - it could simply be HP throwing trademarks at the wall (big companies do that) and tossing some fake products into NDAed catalogs to see which ones leak their products.
We have no idea what hardware it's using, we have no idea what software it's using. We don't even know if they mean "prime" as in "ultimate" (and therefore replacing or being above the 50g), or "prime" as in a "primer", or first graphing calculator (and therefore replacing the 9g).
Here's my pure speculation - please do not take this as fact, this is almost entirely unsourced, just guesses - about this calculator.
First off, I'm guessing that "Prime" does mean "ultimate". So, we're talking a 50g-class machine. The remainder of this post's guesses are based on that.
I suspect the software stack will be based on that of the 39gII, due to Tim Wessman stating that the 39gII's software is the foundation of future high-end calculators. Therefore, this isn't natively a stack-based machine as far as I know, and has no connection to any SysRPL machine other than sharing some user interface conventions, and I believe Tim's said that it has a C rewrite of the 48-series math libraries. I believe the primary programming model will be identical to that of the 39gII, so not at all RPN-like. It may be possible to have backwards compatibility with UserRPL, but I'm unsure if they'll do that or not (and I suspect it'd be difficult).
I do believe that HP will have a form of RPN input, and I suspect the stack will behave like that of an RPL calculator (infinite stack, with Enter confirming entry to the bottom of the stack - the first I'm almost certain of, the second I'm 100% certain of given that Entry RPN exists and is used by all new 4-level RPN calcs).
As for the hardware, I suspect it'll be reminiscent of the 39gII, but with higher build quality (the 39gII has been reported as not being up to the standards of the 50g, as it's marketed as an entry-level student graphing calculator, and was designed as such). System on chip may be completely different - I believe HP's made something portable enough to be independent of that. I haven't dug deep enough into the binary, but I have a suspicion that there's not any ARM code in the emulator, and that it's a straight Win32 build of the 39gII software. If I'm right, that means that they've got freedom to change SoCs - or even CPU architectures for whatever reason - for each calculator that they release, and not break things.
Display, I think they'll stick with greyscale for power consumption reasons. Maybe better than 256x127 (it looks rather bad when a TI-84+ (even if it is the CSE) is beating you on pixel count), but greyscale nonetheless.