One of the Julia creators did a fractal demo [1] recently, including the Sierpinski triangle. Julia is now quite nice for interactive work via the IJulia notebook (IPython kernel for Julia). As a bonus, it does not require a C compiler to get good performance.
IPython is catching up with the Mathematica frontend, slowly but surely. And that Julia demo is nice -- though I don't think a procedural language like MATLAB is the best model for a scientific language in this day and age.
We (Wolfram) are working on a web frontend for the Wolfram Language that will make the language "effectively" open, even if it is still proprietary.
And after the next version, we're going to be concentrating hard on defining a typed subset of the language that can compile via LLVM to all kinds of targets, including the browser.
It's tough to catch up with Mathematica when you guys keep putting out such awesome stuff with each new version. It's been great to see Wolfram innovate so much – versions 6, 7, 8 and 9 were all amazing releases (especially after the four-year gap following 5). I found the new symbolic discrete calculus work in 7 and the symbolic statistics capabilities that built on that in 8 to be particularly mind blowing [1][2]. This seems like it was the fruit of a focused, long-term R&D agenda.
Regarding Julia, although the syntax is superficially similar to Matlab and we've intentionally made many things compatible, Julia isn't really modeled on Matlab. Julia is much more influenced by Lisp – much like Mathematica via Macsyma and Maclisp. Julia's dynamic multiple dispatch is somewhat reminiscent of Mathematica's pattern matching dispatch system too, although, of course, the evaluation semantics are radically different.
I hate to quibble, but having a web frontend for Mathematica that you don't charge people to use does not make it "effectively" open. Mathematica is a shining example of excellence in closed-source software and I have no problem with it as such, but calling it "open" because there is a web version is just disingenuous b.s.
[1] http://nbviewer.ipython.org/url/beowulf.csail.mit.edu/18.337...