Hopefully no one takes this as a flame...
I have both engines and both engines have different strengths and weaknesses. I think in the long run, Torque3D will hands down be the better choice. I only say this because, Torque3D development is really starting to accelerate. They have to. They have lost SO much market share to Unity, that they are starting to give extra perks to licensees. For example, the original reason I stopped using Torque3D and only Leadwerks was because the forest editor was going to cost an extra $300-$500. They just made it free. It has a lot more features that Leadwerks vegetation system, like collideable vegetation. They both use deferred rendering, which I think LE wins at. However, it is optional in Torque3D. If you add the extra baked global illumination, it looks heads and shoulders better than LE. Networking isn't even close. You get source code with Torque3D (which may not be a good thing if you need to make major changes). They are cleaning up the source. They just cleaned up and modularized the physics so it is easy to implement whatever library you want to use.
I think if you are aiming to make a multiplayer game, at this point Torque3D is the better choice. But, if you are limited by $$$$, then LE may be the choice. In the end, I think that LE will have a very difficult time competing with Torque3D and Unity. Not because it isn't as good or better than the other two, but the other two have millions of dollars being invested in them. Competition drives growth and I see both of those competing against each other driving the tech way out of the reach of LE, C4, and the like.
I like LE and plan on doing a project with it, long term I will most likely be using Torque. Hopefully Josh can find some good partners or investors. There is absolutely no way he will compete in the long run as a one man show.