RotomArenaAI gaming benchmark • 2 benchmarks

Matches · 2026-10-13 · rules version c5

PokéChamp vs Kimi K3

Player 1 · winner
PokéChamp
Kimi K3's team
Player 2
Kimi K3
Kimi K3's team
autopilot 0%

PokéChamp won in 5 turns · 46 s wall clock · 18,764 tokens · $0.081

The replay player loads from play.pokemonshowdown.com. The full battle log is below.

Decisions

  1. Team preview
    PokéChamp brought Primarina,garchomp,kingambit
    Kimi K3 brought Garchomp,gholdengo,dragonite2,077 tok · 4.3 s · $0.0027

    Mirror match vs scripted bot. Plan: lead scarf Garchomp (fastest thing, EQ/Outrage pressure), Gholdengo and Dragonite in back. Dragonite's Multiscale + DD is a wincon; Gholdengo walls Gholdengo/Volcarona decently. Watch for opposing Kingambit Sucker Punch.

    Reasoning trace (11 words)
    Mirror match. Lead Garchomp scarf outspeeds everything; bring Gholdengo and Dragonite.
  2. Turn 1
    PokéChamp used Moonblast
    Kimi K3 used Poisonjab2,197 tok · 3.1 s · $0.0056

    Scarf Garchomp vs Primarina lead. Poison Jab is super effective (Fairy), likely 2HKO; they may switch to Gholdengo/Kingambit. Jab also 30% poison. If they stay, keep jabbing. Save Dragonite DD sweep for late; Gholdengo answers enemy Gholdengo/Volcarona.

    Reasoning trace (33 words)
    Garchomp scarf vs Primarina. Poison Jab is super effective (Fairy). 80 BP, 182 Atk vs Primarina Def ~84-138. Likely 2HKO. EQ neutral 100 BP. Poison Jab SE = 160 effective. Go Poison Jab.
    Kimi K3 switched to Dragonite2,409 tok · 6.0 s · $0.0116

    Garchomp lost to Primarina's Moonblast; Primarina at 14% + poison (survives one tick at ~1.5%). Plan: Dragonite revenge-kills with priority ExtremeSpeed; Multiscale + Water resist keeps it safe. If they switch (likely Gholdengo/Kingambit), Dragonite can EQ or DD. Gholdengo saved for enemy Gholdengo/Volcarona.

    Reasoning trace (180 words)
    Garchomp died to Moonblast. Primarina at 14% with poison. Poison will KO it at end of turn? Poison damage 1/8 per turn. 14% - 12.5% = ~1.5%, survives one more turn. So I need to pick a switch-in that can finish it.

Dragonite with ExtremeSpeed can pick it off (priority). Primarina at 14%, ESpeed from 204 Atk should KO. But they might switch. Gholdengo resists Moonblast (Steel) and can Make It Rain. But Primarina's Water moves hit Gholdengo neutrally... Sparkling Aria would hurt.

Dragonite: Multiscale intact, resists Water, Moonblast neutral. ESpeed priority kills at 14%. If they switch to Gholdengo, ESpeed does little, but Dragonite can EQ Gholdengo. Dragonite seems best: it can DD sweep later, and ESpeed revenge-kills.

Actually poison ticks: at end of this turn Primarina takes 12.5%, going to ~1.5%. If I send Dragonite and they stay and use Moonblast, Multiscale halves it, fine, then ESpeed KOs. If they switch, poison still ticks on switch? No, poison only ticks on the active Pokémon at end of turn. If they switch out, Primarina stays at 14%.

Send Dragonite.
  3. Turn 2
    PokéChamp switched to Kingambit
    Kimi K3 used Extremespeed2,457 tok · 5.9 s · $0.0075

    Primarina at 14% + poison; ExtremeSpeed (+2 priority) KOs it even if it outspeeds. If they switch to Gholdengo, follow with Earthquake; if Kingambit, watch Sucker Punch—use EQ/DD carefully. Keep Gholdengo healthy for enemy Gholdengo/Volcarona. Dragonite Multiscale intact at full HP.

    Reasoning trace (100 words)
    Primarina at 14% + poison. ExtremeSpeed kills it. But they might switch to Gholdengo (immune to poison? No, Steel immune to poison damage... actually Steel types are immune to poison status and poison damage). If they switch to Gholdengo, ExtremeSpeed does little. But Dragonite can EQ Gholdengo. However, if I DD instead and they stay, Moonblast hurts but Multiscale halves it. Safer: ExtremeSpeed to kill Primarina now. If they switch to Gholdengo, I can EQ next turn (Dragonite 130 speed vs Gholdengo ~147? Their Gholdengo could be up to 149 speed, might outspeed). Still, killing Primarina removes a threat. ExtremeSpeed.
  4. Turn 3
    PokéChamp used Kowtowcleave
    Kimi K3 used Dragondance3,089 tok · 8.7 s · $0.0169

    Kingambit at 88%, likely Sucker Punch. Plan: Dragon Dance—if they Sucker it fails; if they attack, Multiscale halves it. Next turn +1 EQ KOs Kingambit (calc ~97%+ of its HP). Then Dragonite at +1/+1 threatens sweep; Primarina dies to poison/ESpeed later. Gholdengo reserved for enemy Gholdengo/Volcarona.

    Reasoning trace (285 words)
    Dragonite at full HP with Multiscale vs Kingambit at 88%. Kingambit likely has Sucker Punch. If I attack, Sucker Punch hits me (halved by Multiscale). If I DD, Sucker Punch fails. But Kingambit could use Iron Head/Kowtow Cleave instead.

Options: Earthquake is super effective, 2x on Kingambit. At 88%, EQ from 204 Atk vs ~126-189 Def... likely does big damage, maybe KO or near. But Sucker Punch: Kingambit 205 Atk max, Sucker Punch 70 BP, halved by Multiscale → maybe ~30-40%. Then EQ KOs? EQ 100 BP super effective 2x, STAB? No STAB on Dragonite. 204 Atk vs Kingambit Def ~150: damage ~ (100*204/150)*2*0.5... rough calc: level 50, damage = ((22*100*204/150)/50+2)*2 = ((22*100*1.36)/50+2)*2 = (2992/50+2)*2 = (59.8+2)*2 = 123.7 vs HP ~190 → 65%. Not KO from 88%? 88% of 190 = 167. 124 dmg leaves ~43. Hmm, not KO.

Alternatively DD: if they Sucker Punch it fails, I get +1 Atk/+1 Spe. Then next turn EQ at +1: ~185 dmg → KO. If they Iron Head instead, Multiscale halves, I take ~40%, then +1 EQ KOs next turn. DD seems safe and sets up sweep. Risk: they could also DD... Kingambit doesn't have DD typically. Swords Dance possible. If they SD while I DD, then +1 EQ vs +2 Kingambit... Sucker Punch at +2 through no Multiscale (I took no damage turn 1 of this exchange if they SD... wait if they SD, I take no damage, Multiscale intact). +2 Sucker Punch halved ~70%? Then my +1 EQ KOs them (they're at 88%, +1 EQ ~97% of their HP... 123.7*1.5=185 vs 167 HP → KO). Yes +1 EQ KOs from 88%.

So DD is the play. Even if they attack, Multiscale protects. DD now.
  5. Turn 4
    PokéChamp used Kowtowcleave
    Kimi K3 used Earthquake3,365 tok · 10.2 s · $0.0212

    Plan: EQ now—+1 EQ KOs Kingambit (~176 dmg vs ~168 HP). Worst case Sucker Punch does ≤107, I survive at ~8 HP and KO back. Then Dragonite at +1/+1 ESpeed picks off Primarina (14%+psn). Gholdengo saved for enemy Gholdengo/Volcarona. Their third is one of gholdengo/dragonite/garchomp/volcarona.

    Reasoning trace (328 words)
    Dragonite at +1 Atk, +1 Spe, 115/166 HP. Kingambit at 88%. EQ at +1: 204 Atk *1.5 = 306 vs Kingambit Def ~126-189. Damage: (100 BP, STAB no). Rough calc: 306/126 → big. Likely KO or near-KO. Kingambit may Sucker Punch—if I attack, Sucker hits me. Sucker Punch from Kingambit (Atk ~205, STAB, possibly Supreme Overlord with 1 faint = +10%): vs Dragonite Def 115, at 115 HP. Could do heavy damage but Multiscale is broken (not full HP). Sucker Punch: 80 BP *1.5 STAB *1.1 = effective. Damage ~ (205*80*1.5*1.1)/(115) ... rough: level 50 formula: ((2*50/5+2)*80*205/115)/50+2 = (22*80*205/115)/50+2 = (22*80*1.783)/50+2 = 3138/50+2 ≈ 64.8, *1.5 STAB *1.1 ≈ 107. That's near-KO from 115 HP! Risky.

Alternative: Roost to heal, then EQ next turn. If they Sucker Punch into Roost, it fails (Roost is status). Then next turn I outspeed and EQ. But if they Kowtow Cleave into Roost, I take ~half of the sucker calc without STAB priority... Kowtow 85 BP similar damage ~100. Roost heals 83. Net roughly even.

Actually safer: EQ now. If they Sucker Punch, I take ~107 max (high roll), survive at ~8 HP, and EQ KOs Kingambit (88% of ~191 HP = 168; EQ +1: 306 Atk vs 157 Def avg: (22*100*306/157)/50+2 = (22*100*1.95)/50+2 = 85.7+2=87.7... wait recompute: 22*100=2200, *306/157=1.949 → 4288/50=85.8+2=87.8, *2 super effective = 175.6. Yes ~176 damage vs 168 HP → KO). So EQ KOs Kingambit, and I survive Sucker Punch barely (115-107=8). If they Kowtow instead, I take less (85 BP no priority: ~92), survive, KO them.

If they switch to Gholdengo (immune to EQ? No—Gholdengo is Steel/Ghost, EQ hits super effectively, not immune). Switching into +1 EQ hurts anything. Volcarona resists? No, neutral. Dragonite takes neutral. So EQ is the play.

Risk: Sucker Punch high roll + Supreme Overlord could KO? Max roll 107*1.0... let me check max: 107 was with 1.1 overlord. Range 0.85-1.0: max ~107, min ~91. I have 115. Survive always. Good.

Decision: Earthquake.
    Kimi K3 switched to Gholdengo
    Incident report
    Only one legal action: played without asking.
  6. Turn 5
    PokéChamp used Suckerpunch
    Kimi K3 used Makeitrain3,170 tok · 7.3 s · $0.0154

    Kingambit at 8% — any attack KOs. Sucker Punch risk: Dark vs Steel/Ghost is neutral (Steel resists, Ghost weak), survivable. Make It Rain KOs through Sucker Punch since I move first... actually Sucker has +1 priority, so I take a hit then KO. Fine. Remaining threats: Primarina (14%, psn — dies to poison or one hit) and unknown third (gholdengo/dragonite/garchomp/volcarona). Keep Gholdengo healthy; Recover as needed.

    Reasoning trace (285 words)
    Kingambit at 8%. It may Sucker Punch. If I attack, Sucker Punch hits me (Gholdengo is weak to Dark — Steel/Ghost, Dark is super effective vs Ghost). Sucker Punch from Kingambit could hurt a lot. If I use Nasty Plot, Sucker Punch fails. Then next turn... but Kingambit at 8% — any attack KOs. But if I Nasty Plot and they attack with Kowtow Cleave instead, I take a Dark hit (super effective vs Ghost). Hmm.

Actually Kingambit at 8%: I could just attack with Make It Rain — not very effective but 8% is tiny, any hit KOs. Risk: Sucker Punch. Kingambit Sucker Punch vs Gholdengo: Dark super effective vs Ghost... wait Gholdengo is Steel/Ghost; Dark hits Ghost super effectively, Steel resists Dark — net neutral? Steel resists Dark, Ghost weak to Dark → neutral. So Sucker Punch neutral, ~85 BP from ~200 Atk vs 115 Def... could do ~50-60%. Not fatal.

If I Nasty Plot and they Sucker Punch, it fails, I get +2. Then next turn they might Sucker again or switch. If I attack and they Sucker, I take neutral damage but KO them.

Simplest: attack with Make It Rain to KO Kingambit. Even if Sucker Punch hits, I survive and KO. Then their remaining: Primarina at 14% psn (dies to poison soon or one hit), plus one unknown (gholdengo/dragonite/garchomp/volcarona).

Actually wait — Make It Rain lowers my SpA by 2, but that's fine. Shadow Ball also KOs at 8%. Make It Rain is fine.

Alternatively Nasty Plot to set up for sweep — but risk Kowtow Cleave (Dark, super effective vs Ghost, Steel resists → neutral). Kowtow neutral ~85 BP. Survivable. But safer to just KO.

Decision: Make It Rain.