[WiSaGaN]

If what here says is true: https://x.com/teortaxesTex/status/1880768996225769738, then R1 may as well just be the better model. You can scale up R1 with lower token count to achieve better than o1 high results.