A baby dragon would be amazing. I'd love to see a 'serpent-style' dragon without legs or maybe with only vestigial legs (imagine it learning to 'walk'!) and something with wings as well. You could express a lot with wings!
As irbyma7297 pointed out, some of these things are really difficult to construct - some of you may have heard of Honda's ASIMO robot, which was millions of dollars in the making and still had an awkward habit of falling down stairs.

I suspect one of the other problems is just that making robots like Pleo is problematic, both in the difficulty of marrying technology and cuddly exteriors (like the skin-splitting problems with Ugobe Pleo) when you can get away with making something more traditionally robotic like an Aibo that people might even see as less of a 'toy' and be willing to spend more on, and in people being frankly suspicious of robots that are lovable. I suspect it may just be financially more sensible for Innvo to stick with the Pleo they have, or not to try too hard to come up with anything else. (I'd love to be wrong and find they've been coming up with a totally new adorable creature!)
I'm very much of the opinion though that the only way we're going to get more life forms of this kind, and especially ones we can tinker with and program ourselves, is to get "ordinary" people who are passionate about the idea and have the ability to build robots to get together and do the hard work of developing the technology themselves. If someone was able to come up with a robot prototype on their own time that worked well, looked cute and so on, in these days of the Internet they could use something like Kickstarter to finance getting it built professionally - so long as they didn't make the mistake so many others have of collecting money for a product they couldn't provide in a timely manner or at all...! Developers who did a lot of the initial legwork in their spare time for the love of it might also be able to keep the cost down a bit (I think Pleo's cost is reasonable given its sophistication, but not everyone will see that or care) and would perhaps be more willing to allow the public to have access to all the details of the design, source code and so on. Companies generally want to hold onto this 'intellectual property' and would be under pressure to do so from their investors. Pleo is adorable all on its own without us needing to have all the tools the developers had to program it, but I do think consumers will need full control over the programming of their robots, if they want that, for the idea to really capture the wider public's imagination.
As you might be able to tell from the length of my rantings on this, building and selling my own robots is something I'd love to try in the future, although it's well beyond my abilities now. I'm keen on learning, though, and also showing Pleo off to people I know socially who have the abilities and equipment to build robots but have maybe never had their thoughts wander in that direction.