Testrun Stockfish-26-07-26, 27.000 games, incredible progress

[Rebel]

Code: Select all

STC Rating List       : 2026-07-22
Time Control          : Blitz 40m/10s single thread.
Software              : cutechess-cli
CPU                   : AMD Ryzen 9 7950X3D 16-Core
GPU                   : RTX 4080 SUPER

Code: Select all

   # PLAYER                   :  RATING  ERROR   POINTS  PLAYED   (%)  CFS(%)      W      D      L  D(%)
   1 Stockfish-26-07-26       :  3823.8    2.9   5823.5    9189    63     100   2817   6013    359    65
   2 Leela-0.32.1-BT4         :  3790.9    5.0   3154.5    6000    53      98    996   4317    687    72
   3 Leela-0.32-BT3           :  3787.2    3.6  20697.5   33000    63      90  10284  20827   1889    63
   4 Stockfish-2026-01-11     :  3784.5    2.7  26612.0   41778    64      52  13712  25800   2266    62
   5 Stockfish-18             :  3784.4    1.7  40106.0   66000    61     100  18783  42646   4571    65
   6 Reckless-0.10.0-dev      :  3774.3    2.6  16936.0   28022    60     100   7878  18116   2028    65
   7 Reckless-0.90-final      :  3768.7    2.0  34119.5   57000    60      62  15394  37451   4155    66
   8 Stockfish-TI             :  3768.1    3.0  18987.5   30000    63     100   9635  18705   1660    62
   9 PlentyChess-8.0          :  3759.7    2.6  22822.0   40020    57      64   9284  27076   3660    68
  10 Stockfish-17.1           :  3759.0    2.0  28119.5   48000    59      78  12550  31139   4311    65

After 9.189 of 27.000 games an elo gain of 39 !

[Rebel]

Code: Select all

STC Rating List       : 2026-07-28
Time Control          : Blitz 40m/10s single thread.
Software              : cutechess-cli
CPU                   : AMD Ryzen 9 7950X3D 16-Core
GPU                   : RTX 4080 SUPER

Code: Select all

   # PLAYER                   :  RATING  ERROR   POINTS  PLAYED     W      D      L  D(%)
   1 Stockfish-26-07-26       :  3824.5    2.6  17135.0   27000  8336  17598   1066    65
   2 Leela-0.32.1-BT4         :  3790.9    4.6   3154.5    6000   996   4317    687    72
   3 Reckless-0.10.0-dev      :  3773.9    2.1  17776.5   30000  7999  19555   2446    65
   4 PlentyChess-8.0          :  3759.5    2.0  23628.5   42000  9392  28473   4135    68
   5 Pawnocchio-2.0.1         :  3729.2    2.3  20106.5   39000  6613  26987   5400    69
   6 Viridithas-20.0.0        :  3721.1    1.3  17923.5   36000  5731  24385   5884    68
   7 Coda-0.9.3-AI            :  3708.4    3.1  14520.0   30000  4545  19950   5505    67
   8 Alexandria-9.0           :  3706.6    1.9  34109.0   67865 11136  45946  10783    68
   9 Obsidian-16              :  3705.3    1.5  58519.5  116652 19176  78687  18789    67
  10 Stormphrax-8.0.0         :  3695.2    1.1  20946.5   42000  6878  28137   6985    67

Stockfish-26-07-26 : 3824
Stockfish-18 : 3784 +40 elo

. Stockfish finally on top where it belongs.
. Leela lost first place despite considerable better hardware.
. Elo gap (not counting Leela) is +50 elo once gain with the runner up.
. Will the relative newcomers Reckless and PlentyChess give up?

Congrats to Stockfish team.