From Speedsolving.com Wiki
Proposer(s): Rowe Hessler, Chester Lian
Proposed: SV: 2009
The rest of HLS: 2013
Alt Names:
Variants: SV, UF, UL, UR, UFUR, UFUL, ULUR, and all edges misoriented
Subgroup: unknown
No. Algs: 432, including mirrors
Avg Moves:
Previous state: HLS setup cube case
Next state: PLL

HLS setup cube case -> HLS step -> PLL

The HLS step is the step between the HLS setup cube case and the PLL.

HLS, short for Hessler Last Slot, is a subset of OLS. The HLS substep solves the last F2L pair, if the edge and corner can be paired with one move and can be paired and inserted with three moves, and it also skips OLL which is the third step used in the widely popular CFOP method. HLS can be used in speedsolving or FMC to decrease move count.

HLS was partially developed and shared publicly in 2009 by Chester Lian in the subset Summer Variation, or SV. The idea of the rest of HLS was shared publicly in 2013 by Rowe Hessler, however the algorithms have never been completed apart from SV.

Learning Approach

There are 8 subsets under HLS. They are named after which edges are misoriented if the last F2L pair's corner and edge pieces are set up in this way (Setup: RU'R'). Note that there are mirrored cases which still fall under their unmirrored subsets. These subsets include SV, UF, UL, UR, UFUR, UFUL, ULUR, and all edges. Each of these subsets include 54 algorithms, if mirrors are counted as separate cases. It is usually recommended to first learn SV and then the rest of HLS in whatever order you choose. Learn the placement of the top layer's stickers, like they are in OLL.


  • Move count is decreased compared to normally doing the last F2L pair, then OLL.
  • It requires less look ahead, if implemented into solves, compared to doing the last F2L pair and OLL.
  • You'll instantly gain cool kid points and be able to impress your cubing friends


  • There are a total of 432 algorithms, including mirrors.
  • Because of the first point, this means that if the solver were to learn full HLS, it would likely take at least a year if 1 algorithm was learned per day.

See Also

External Links