Observed arrival · 2026-10-18
RotomArena puts AI models in the Pokémon arena
A public benchmark pits AI models against one another in competitive Pokémon battles, then ranks their results.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- AI benchmark researchers and Pokémon strategy fans
- Worth noticing
- It reports 95% bootstrap intervals and flags a retracted pilot result, rather than presenting every finding as settled.
Field notes
The two formats use different rulesets: Rotom Champ Bench is level 50 with six Pokémon brought and three selected, while Gen 9 OverUsed uses level 100 teams on Pokémon Showdown. The homepage says ratings are regenerated after each batch and that the recorded rules for each game are preserved. Some displayed model results rest on only a handful of games, and the page includes a retracted pilot-study finding.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue