BET applies to type 2 and 4. That is basically unanimous.
The sample that shows loop of H3 type have (very likely) no micropores (small uptake in low p/p0 range). In this case you can apply BET theory in common p/p0 range (0.05-0.3). The sample that shows H4 hysteresis have micropores (large uptake in low p/p0 range). The BET theory can be applied here, but in a much lower p/p0 range. For activated carbon, for example, one would look for a linear(!) range between p/p0 = 0.02-0.1. Generally, the BET theory is not suitable for microporous materials.