gtp4zen except 19x19 works with boards 9x9, 13x13 and 11x11 in sabaki and gogui
gtp4zen 0.3.5 https://github.com/breakwa11/gtp4zen/releases

network 13x13 (b) obtained from lm gx89 (https://github.com/leela-zero/leela-zero/issues/2240 by ChangeBoardSizeOfWeight.cpp) against gtp4zen (w): 48-2.
cpuonly, threads 4, 10 sec/move, komi 7.5, gogui-twogtp 1.5.1
gtp4zen zen7 9dan -r20 -s6000 -n3 -o1.0 -p0.75 (https://github.com/breakwa11/GoAIRating ... /AIcmds.md)

Some 9x9 games between

katago 20b (Sabaki)
Crazy Stone DL (PC, set at 7 dan)
gtp4zen7 (Sabaki, -z 7 -t 4 -T 5 -s 200000)
Lazy Baduk (default settings , Huawei tablet)
gnugo (Sabaki)

komi 7.5
gnugo plays instantly, around 4-5 sec/move for the others.

Katago-CSDL 4-0
Katago-Zen 4-0
Katago-LB 4-0
CSDL-Zen 3-1 (one win as W for Zen)
CSDL-LB 4-0 (but close games)
LB-Gnugo 4-0
LB-Zen 4-0
Zen-Gnugo 4-0

50 game match : katago 20b v gtp4zen7
9x9, 5sec/move, komi 7.5, twogtp, 9 duplicate games, 7 games without resignation, 0 error, 1 game where both programs don't agree on the score.
Average game length 49, average time Zen : 82s, average time Katago 114s
Katago wins 41-0
Maybe Zen parameters not optimal, I don't know.

Vargo, KataGo when playing on equal terms, it is much stronger than Zen. Does KataGo use a video card for the game? then he has even more advantage!

another 50 games lm (w)-zen (b) (13x13) 41-9. after 100 games 89-11

I was asked a question in pm about zen 7 settings. The original zen 7: 6 kyu - 9 dan, or the time per move 5 sec - 120 sec.The original one is not able to play directly with another program through the gtp. development of third-party programmers uses Zen.dll and allows play through the gtp: https://github.com/breakwa11/gtp4zen/releases
settings:
https://lifein19x19.com/viewtopic.php?p=241847#p241847
https://lifein19x19.com/viewtopic.php?p=241850#p241850