This depends on how much mass there is, whether it's concentrated in a small lump, or flows through the paths of least resistance, and separates, and spreads out, (like chernobyl did). If it spreads out, the reaction slows, and then, it's largely decay-heat that's left over (which is pretty significant, but still, not 3000 degrees C significant).
Youtube is full of videos of steam-clouds that have been around the plant since roughly June 2012. This *seems* to indicate that there's something hot contacting moisture in the ground, and forcing out steam over an area of soil, but it's not consistent with a 77-ton mass of molten material dropping into a lake (ie. worst-case steam-explosion situation which is often portrayed as the result of a meltdown). The plant is along the coast, so fog has never been an unusual occurrence, so that likely conceals the fact that the core is issuing radioactive steam through cracks in the soil. A simple sampling and test of the steam should reveal if this were the case. But we're not getting that kind of data from Tepco.