RotomArenaAI gaming benchmark • 2 benchmarks

Retracted · 2026-08-30 · rules version v0

Nemotron_f6e vs Gemini 3.7_9c3

Retracted. This match was voided: nemotron_3_ultra_550b_a55b was on autopilot for 61% of its turns. It is kept for transparency and does not count toward any rating. Why we retracted the pilot study.
Player 1
Nemotron_f6e
team not recorded
autopilot 61%
Player 2 · winner
Gemini 3.7_9c3
team not recorded
autopilot 9%

Gemini 3.7_9c3 won in 18 turns · 14 min wall clock · 64,573 tokens · $0.041

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

Decisions

  1. Turn 1
    Gemini 3.7_9c3 used Closecombat1,768 tok · 13.9 s · $0.0017

    Iron Valiant outspeeds Ting-Lu and threatens a 4x effective STAB move. Vessel of Ruin mitigates special attacks, making physical Close Combat the highest-damage option against Ting-Lu or whatever switches in.

    Nemotron_f6e used Stealthrock2,306 tok · 79.3 s

    Iron Valiant (Fairy/Fighting) is immune to Earthquake and resists Ruination. Ting-Lu outspeeds (126 vs 116) and has massive bulk to set hazards safely. Stealth Rock is the optimal turn 1 play to chip switch-ins and limit Valiant's longevity.

  2. Turn 2
    Gemini 3.7_9c3 used Closecombat1,765 tok · 10.4 s · $0.0016

    Ting-Lu is at 60% HP. Close Combat easily finishes off Ting-Lu, removing a massive defensive pillar.

    Nemotron_f6e used Earthquake1,853 tok · 44.3 s

    Iron Valiant has -1 Def and -1 Spe, making it slower and physically frail. Ting-Lu's Earthquake is super effective (Ground vs Fairy/Fighting) and will likely OHKO. No need to Tera or switch; capitalize on the stat drops immediately.

  3. Turn 3
    Gemini 3.7_9c3 used Moonblast1,788 tok · 13.9 s · $0.0017

    Iron Valiant easily outspeeds Ting-Lu (45 base Speed vs 364 Speed). At 15% HP, Moonblast is a guaranteed knockout without incurring further stat drops.

    Nemotron_f6e used Earthquake1,847 tok · 46.6 s

    Iron Valiant at 6% HP with -2 Def/Spd. Earthquake is super effective Ground vs Fairy/Fighting and guarantees OHKO. No Tera needed. Ruination only halves. Finish it now.

    Nemotron_f6e switched to Kingambit1,815 tok · 49.8 s

    Iron Valiant at 6% HP with -2 Def/Spd. Kingambit resists both STABs, has priority Sucker Punch for guaranteed KO, and Supreme Overlord boosts power. Best switch-in to finish safely.

  4. Turn 4
    Gemini 3.7_9c3 used Closecombat1,915 tok · 14.5 s · $0.0019

    Iron Valiant is at low HP and faces a 4x weakness to Fighting on Kingambit. Kingambit likely clicks Sucker Punch or Iron Head to revenge kill, but Close Combat decisively OHKOs Kingambit if it does not Sucker Punch or if Valiant trades/attacks. Using Close Combat directly is optimal.

    Nemotron_f6e used Suckerpunch1,851 tok · 49.2 s

    Iron Valiant at 6% HP with -2 Def/Spd. Sucker Punch priority OHKOs reliably. No need to Tera or predict switches.

    Gemini 3.7_9c3 switched to Ogerpon1,743 tok · 14.2 s · $0.0017

    Iron Valiant fainted. Ogerpon-Wellspring is the cleanest offensive check to Kingambit, comfortably taking Sucker Punch and immediately applying tremendous pressure with Ivy Cudgel or Swords Dance.

  5. Turn 5
    Gemini 3.7_9c3 used Ivycudgeltera1,875 tok · 14.3 s · $0.0018

    Ogerpon-Wellspring outspeeds Kingambit. Ivy Cudgel (Water-type via Wellspring Mask) deals heavy neutral damage. Tera Water activates Embody Aspect (+1 SpD) and boosts Ivy Cudgel to safely secure massive damage without risk.

    Nemotron_f6e switched to Great3,015 tok · 84.2 s

    Ogerpon-W threatens Kingambit with 4x Superpower. Great Tusk outspeeds, resists Fighting, hits super effectively with Ice Spinner, and removes Stealth Rock via Rapid Spin. Switching gains momentum and hazard control.

    Nemotron_f6e switched to Kingambit1,884 tok · 51.4 s

    Ogerpon-Wellspring at 88% HP, Tera Water, +1 SpD. Ivy Cudgel is Grass-type. Kingambit resists Grass (Dark/Steel), threatens OHKO with Kowtow Cleave, and has Sucker Punch priority. Best defensive and offensive switch-in.

  6. Turn 6
    Gemini 3.7_9c3 used Ivycudgel1,879 tok · 14.1 s · $0.0017

    <decision> ACTION: MOVE: ivycudgel </decision>

    Nemotron_f6e used Ironhead2,216 tok · 50.0 s

    Ogerpon-W (Water Tera) at 88% HP with +1 SpD. Kingambit resists Ivy Cudgel, outspeeds, and Iron Head (Steel STAB) hits super effectively for likely 2HKO. Attack now to pressure before it boosts further or switches.

  7. Turn 7
    Gemini 3.7_9c3 used Ivycudgel1,896 tok · 14.3 s · $0.0017
    Nemotron_f6e used Kowtowcleave2,617 tok · 61.2 s

    Ogerpon-Wellspring at 49% HP, Water Tera, +1 SpD. Kingambit's Kowtowcleave (85 BP Dark, STAB, Supreme Overlord +20% Atk from 2 fainted allies) OHKOs through neutral bulk. Priority Sucker Punch is weaker and prediction-reliant. Switching loses momentum; Kingambit secures the KO now.

    Nemotron_f6e switched to Dragapultautopilot
    Incident report
    [TIMEOUT] Turn calculation exceeded limit (83.9s). Automated move executed.
  8. Turn 8
    Nemotron_f6e used Shadowballautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 switched to Kingambit2,025 tok · 14.6 s · $0.0019

    Dragapult outspeeds Ogerpon-Wellspring even at +1 SpD and threatens an immediate KO with Dragon Darts or Shadow Ball. Switching directly into Kingambit safely absorbs Ghost and Dragon moves due to its typing and Supreme Overlord boosts, establishing immediate offensive pressure.

  9. Turn 9
    Nemotron_f6e used Shadowballautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Kowtowcleave1,964 tok · 9.3 s · $0.0017

    Nemotron_f6e switched to Gholdengoautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
  10. Turn 10
    Nemotron_f6e used Shadowballautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Kowtowcleave2,063 tok · 14.2 s · $0.0018

    Gemini 3.7_9c3 switched to Ribombeehallucination1,952 tok · 14.3 s · $0.0016
    Incident report
    [FALLBACK] LLM proposed invalid action 'MOVE: '. Chose default valid move. 
Original LLM Reasoning:
  11. Turn 11
    Nemotron_f6e used Shadowballautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Uturn2,140 tok · 14.1 s · $0.0021

    Opponent has Gholdengo and one unrevealed Pokémon left. Gholdengo has Good as Gold (immune to status/Sticky Web/Stun Spore). U-turn breaks Air Balloon and safely pivots into Choice Scarf Gholdengo to easily outspeed and KO with Shadow Ball.

    Gemini 3.7_9c3 switched to Samurott1,985 tok · 14.1 s · $0.0019

    Hisuian Samurott resists Ghost-type attacks from opposing Gholdengo, threatens massive Dark-type damage with Knock Off, and maintains a full HP Focus Sash.

  12. Turn 12
    Nemotron_f6e used Shadowballautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Knockoff2,023 tok · 14.0 s · $0.0017

    <decision> ACTION: MOVE: knockoff </decision>

  13. Turn 13
    Nemotron_f6e used Shadowballautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Suckerpunch2,055 tok · 14.1 s · $0.0018

    Opposing Gholdengo is at 26% HP. Sucker Punch provides guaranteed priority STAB to cleanly KO Gholdengo before it can move.

    Nemotron_f6e switched to Dragoniteautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
  14. Turn 14
    Nemotron_f6e used Extremespeedautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Knockoff2,115 tok · 13.9 s · $0.0018

    <decision> ACTION: MOVE: knockoff </decision>

    Gemini 3.7_9c3 switched to Ogerponhallucination2,022 tok · 14.1 s · $0.0017
    Incident report
    [FALLBACK] LLM proposed invalid action 'MOVE: '. Chose default valid move. 
Original LLM Reasoning:
  15. Turn 15
    Nemotron_f6e used Extremespeedautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Ivycudgel2,118 tok · 13.6 s · $0.0018

    Gemini 3.7_9c3 switched to Gholdengo2,003 tok · 14.6 s · $0.0017

    <decision> ACTION: SWITCH: gholdengo </decision>

  16. Turn 16
    Nemotron_f6e used Extremespeedautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Shadowball2,069 tok · 14.4 s · $0.0017
  17. Turn 17
    Nemotron_f6e used Extremespeedautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Shadowball2,001 tok · 13.6 s · $0.0017

    <decision> ACTION: MOVE: shadowball </decision>

  18. Turn 18
    Nemotron_f6e used Extremespeedautopilot
    Incident report
    [TIME BANK EXHAUSTED] Match time bank depleted. Automated move executed.
    Gemini 3.7_9c3 used Shadowball2,005 tok · 13.2 s · $0.0017

    <decision> ACTION: MOVE: shadowball </decision>