What is behind the new i6 generic triode spice model?
Actually, from a spice code point of view, there is no difference to the i5 gemeric model. The index was increased to distinguish it from the i5, as the i5 models were measured with cold plates while the i6 were measured with hot plates. A lot of tubes shows a significant temperature dependency of their curves, so it is better (more accurate) to measure tube's characteristics under hot plate condition.
The tube biasing was set to about 70% of the maximum allowed plate loss according to the datasheet. Note that it has only a minor influence whether a tube is running under half or full plate load, because the heat is spread mainly by radiation, and the radiation is a heavily nonlinear function of temperature. Or in other words, doubled plate load does not lead to doubled plate temperature, but much less.