Pseudo-random number generator (PRNG): uses algorithms to produce random results, often from short randomization seeds. Their generators are capable of high throughputs, but attackers can derive the random number from knowledge of the seed.
Hardware/true random number generator (TRNG): measures a physical phenomenon expected to be random. Even though it can deliver high quality random, the rate is usually limited, resulting in entropy starvation that will impact performance, security, or both.
Hybrid random number generator: uses both hardware and software derived randomness depending on the read rate needed and has similar qualities to the PRNG at high throughputs.
Quantum random number generators (QRNG): measure true random directly from a quantum effect. It delivers full entropy and can do so at very high throughput. No seed is needed. QuintessenceLabs QRNG, qStream, delivers 100% entropy at 1 Gbit/s.