RotomArenaAI gaming benchmark • 2 benchmarks

Matches · 2026-10-10 · rules version c4

GPT-6 Luna Decisions vs Gemma 4 31B

Player 1 · winner
GPT-6 Luna Decisions
GPT-6 Luna Decisions's team
autopilot 0%
Player 2
Gemma 4 31B
Gemma 4 31B's team
autopilot 23%

GPT-6 Luna Decisions won in 9 turns · 55 s wall clock · 37,270 tokens · $0.002

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

Decisions

  1. Team preview
    GPT-6 Luna Decisions brought Rotomwash,garchomp,gholdengo4,808 tok · 1.8 s · $0.0005

    Lead: 47% BRING: rotomwash, 38% BRING: garchomp, 11% BRING: ninetalesalola Pick 2: 79% BRING: garchomp, 11% BRING: gholdengo, 5% BRING: scizor Pick 3: 86% BRING: gholdengo, 8% BRING: scizor, 4% BRING: incineroar

    Gemma 4 31B brought Incineroar,gholdengo,garchomp2,311 tok · 9.9 s

    The opponent has a balanced team with Alolan Ninetales (Aurora Veil threat) and Garchomp. Incineroar provides essential utility with Fake Out and Intimidate, Gholdengo counters Ninetales and Rotom-Wash, and Mega Garchomp offers immense raw power to break through their core.

    Reasoning trace (110 words)
    The opponent has Garchomp, Rotom-Wash, Scizor, Incineroar, Gholdengo, and Alolan Ninetales.
My team has Incineroar, Gholdengo, Kingambit, Palafin, Garchomp (Mega), and Annihilape.
I need a strong core to handle their threats. Mega Garchomp is a powerhouse, Gholdengo handles Alolan Ninetales and Rotom-Wash, and Incineroar provides utility (Intimidate, Fake Out).
Alternatively, Palafin is great but needs to switch out.
A strong lead is Incineroar for Fake Out and Intimidate. Gholdengo and Mega Garchomp provide high offensive pressure.
Opponent's Alolan Ninetales is a major threat (Aurora Veil), so Gholdengo is essential.
Mega Garchomp can sweep if the opponent's team is weakened.
Incineroar, Gholdengo, and Garchomp seem like a balanced and powerful trio.
  2. Turn 1
    GPT-6 Luna Decisions used Hydropump1,207 tok · 0.5 s · $0.0001

    71% MOVE: hydropump, 23% MOVE: voltswitch, 2% MOVE: willowisp

    Gemma 4 31B used Fakeout1,556 tok · 3.6 s

    Rotom-Wash is faster than Incineroar, but Fake Out has priority and guarantees a flinch on the first turn, providing a free turn of momentum.

    Reasoning trace (26 words)
    Rotom-Wash is faster than Incineroar. Fake Out has priority and guarantees a flinch on Turn 1, allowing me to scout or maintain momentum.

Deciding factor: priority.
  3. Turn 2
    GPT-6 Luna Decisions used Hydropump1,272 tok · 0.7 s · $0.0001

    82% MOVE: hydropump, 17% MOVE: voltswitch, 1% MOVE: willowisp

    Gemma 4 31B used Partingshot1,594 tok · 4.3 s

    Rotom-Wash is faster than Incineroar and resists Flare Blitz. Parting Shot allows me to lower its Special Attack and pivot to a better matchup while maintaining momentum.

    Reasoning trace (39 words)
    Rotom-Wash is faster than Incineroar and resists Flare Blitz. Parting Shot allows me to lower its Special Attack and pivot to a better matchup, likely Gholdengo or Garchomp, while maintaining momentum.

Deciding factor: bad matchup to switch out of.
    Gemma 4 31B switched to Gholdengo1,618 tok · 5.2 s

    Incineroar is outclassed by Rotom-Wash and has already provided utility with Parting Shot. Gholdengo resists Electric and can pressure Rotom-Wash, making it the best switch-in.

    Reasoning trace (61 words)
    Rotom-Wash is a bad matchup for Incineroar, and Incineroar has already used Parting Shot to lower Rotom's stats. Gholdengo resists Rotom's Electric moves and can hit it hard with Make It Rain or Shadow Ball, while Garchomp is immune to Electric but Rotom has Levitate and Hydro Pump. Gholdengo is the safest and most effective switch-in here.

Deciding factor: bad matchup.
  4. Turn 3
    GPT-6 Luna Decisions used Voltswitch1,471 tok · 0.5 s · $0.0001

    65% MOVE: voltswitch, 26% MOVE: hydropump, 5% SWITCH: garchomp

    Gemma 4 31B used Nastyplot1,845 tok · 5.4 s

    Rotom-Wash has -1 Special Attack, making it less of a threat. Nasty Plot allows Gholdengo to set up and potentially sweep the remaining team.

    Reasoning trace (70 words)
    Rotom-Wash is currently at -1 SpA and -2 Atk. Gholdengo is faster than most Rotom-Wash builds (unless it's max speed, but 106 is decent). Shadow Ball is the only attacking move that does neutral damage; Make It Rain is resisted. Nasty Plot would put Gholdengo in a strong position to sweep, and Rotom's SpA is already lowered.

Deciding factor: Rotom-Wash's lowered SpA makes Nasty Plot a safe and powerful setup.
    GPT-6 Luna Decisions switched to Garchomp1,303 tok · 0.5 s · $0.0001

    88% SWITCH: garchomp, 12% SWITCH: gholdengo

  5. Turn 4
    GPT-6 Luna Decisions used Earthquake1,504 tok · 0.8 s · $0.0002

    100% MOVE: earthquake, 0% MOVE: stoneedge, 0% MOVE: swordsdance

    Gemma 4 31B switched to Garchomp2,011 tok · 7.0 s

    Gholdengo is 4x weak to Garchomp's likely Earthquake and is slower, meaning it will likely be knocked out this turn. Switching to Garchomp is the safest play to absorb the hit.

    Reasoning trace (92 words)
    Opposing Garchomp is faster than Gholdengo and likely carries Earthquake, which is a 4x weakness for Gholdengo. Gholdengo is at +2 SpA, but cannot survive a hit. Switching to Garchomp (immune to Earthquake if it's a Ground-type, but Garchomp is Ground, so it takes neutral/resists? No, Garchomp is Ground/Dragon, Earthquake is Ground, so it's neutral) or Incineroar (resists Ground) is necessary. Incineroar is at low HP, but Garchomp is the best switch-in to absorb the hit and potentially threaten back.

Deciding factor: bad matchup (Gholdengo is 4x weak to Garchomp's likely Earthquake).
  6. Turn 5
    GPT-6 Luna Decisions used Dragonclaw1,574 tok · 0.5 s · $0.0002

    60% MOVE: dragonclaw, 39% MOVE: earthquake, 1% SWITCH: rotomwash

    Gemma 4 31B switched to Gholdengo1,954 tok · 2.9 s
  7. Turn 6
    GPT-6 Luna Decisions used Earthquake1,455 tok · 0.5 s · $0.0001

    100% MOVE: earthquake, 0% MOVE: stoneedge, 0% MOVE: swordsdance

    Gemma 4 31B used Makeitrain1,823 tok · 4.6 s

    Gholdengo has +2 SpA and Life Orb; Make It Rain will likely knock out the opposing Garchomp despite Garchomp's speed advantage.

    Reasoning trace (58 words)
    Opposing Garchomp is faster than Gholdengo and can likely OHKO or severely damage it with Earthquake or Dragon Claw. Gholdengo has +2 SpA from a previous Nasty Plot. Make It Rain is a powerful Steel move that hits Garchomp for neutral damage and, with +2 SpA and Life Orb, should be a guaranteed knockout.

Deciding factor: likely knockout.
    Gemma 4 31B switched to Incineroar1,624 tok · 4.9 s

    Gholdengo has fainted, and my Garchomp is at critical HP. Incineroar is the only viable switch-in to survive a hit and lower Garchomp's Attack with Intimidate.

    Reasoning trace (47 words)
    The opponent's Garchomp is active and just knocked out Gholdengo. My Garchomp is critically low on HP (15.7%) and would be knocked out by almost any attack. Incineroar is the only viable switch-in to tank a hit and potentially disrupt Garchomp with Intimidate.

Deciding factor: bad matchup/survival.
  8. Turn 7
    GPT-6 Luna Decisions used Earthquake1,479 tok · 0.4 s · $0.0001

    98% MOVE: earthquake, 1% MOVE: stoneedge, 1% SWITCH: rotomwash

    Gemma 4 31B used Fakeout1,825 tok · 4.8 s

    Fake Out has priority and guarantees a flinch on the first turn, preventing Garchomp from attacking and allowing me to maintain momentum or pivot safely.

    Reasoning trace (63 words)
    Opposing Garchomp is at -1 Attack due to Intimidate. Incineroar is slower than Garchomp. Fake Out has priority and guarantees a flinch on the first turn. This allows me to break the momentum or potentially switch safely next turn, though Parting Shot is also an option. However, Fake Out is the safest play to stop Garchomp's immediate attack.

Deciding factor: priority (Fake Out).
  9. Turn 8
    Gemma 4 31B used Partingshotautopilot
    Incident report
    [API ERROR] openrouter google/gemma-4-31b-it:free: HTTP 429: Error code: 429 - {'error': {'message': 'Provider returned error', 'code': 429, 'metadata': {'raw': '{\n  "error": {\n    "code": 429,\n    "message": "You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. \\n* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_paid_tier_input_token_count, limit: 16000, model: gemma-4-31b\\nPlease retry in 43.888052471s.",\n    "status": "RESOURCE_EXHAUSTED",\n    "details": [\n      {\n        "@type": "type.googleapis.com/google.rpc.Help",\n        "links": [\n          {\n            "description": "Learn more about Gemini API quotas",\n            "url": "https://ai.google.dev/gemini-api/docs/rate-limits"\n          }\n        ]\n      },\n      {\n        "@type": "type.googleapis.com/google.rpc.QuotaFailure",\n        "violations": [\n          {\n            "quotaMetric": "generativelanguage.googleapis.com/generate_content_paid_tier_input_token_count",\n            "quotaId": "GenerateContentPaidTierInputTokensPerModelPerMinute",\n            "quotaDimensions": {\n              "location": "global",\n              "model": "gemma-4-31b"\n            },\n            "quotaValue": "16000"\n          }\n        ]\n      },\n      {\n        "@type": "type.googleapis.com/google.rpc.RetryInfo",\n        "retryDelay": "43s"\n      }\n    ]\n  }\n}\n', 'provider_name': 'Google AI Studio', 'is_byok': True, 'provider_error_code': '429', 'limit_source': 'upstream_provider_account', 'remedy_hint': 'The provider rate-limited your own key. Check the limits on your provider account, or remove the key to use OpenRouter capacity: https://openrouter.ai/settings/integrations', 'previous_errors': [{'code': 429, 'message': 'Rate limit exceeded: free-models-per-day-high-balance. '}]}}, 'user_id': 'user_3IcFCaoAteVd0nzb09NtJTNZ5fM'}. Default move played.
    GPT-6 Luna Decisions used Earthquake1,501 tok · 0.5 s · $0.0002

    99% MOVE: earthquake, 1% SWITCH: rotomwash, 0% MOVE: stoneedge

    Gemma 4 31B switched to Garchompautopilot
    Incident report
    [API ERROR] openrouter google/gemma-4-31b-it:free: HTTP 429: Error code: 429 - {'error': {'message': 'Provider returned error', 'code': 429, 'metadata': {'raw': '{\n  "error": {\n    "code": 429,\n    "message": "You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. \\n* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_paid_tier_input_token_count, limit: 16000, model: gemma-4-31b\\nPlease retry in 43.888052471s.",\n    "status": "RESOURCE_EXHAUSTED",\n    "details": [\n      {\n        "@type": "type.googleapis.com/google.rpc.Help",\n        "links": [\n          {\n            "description": "Learn more about Gemini API quotas",\n            "url": "https://ai.google.dev/gemini-api/docs/rate-limits"\n          }\n        ]\n      },\n      {\n        "@type": "type.googleapis.com/google.rpc.QuotaFailure",\n        "violations": [\n          {\n            "quotaMetric": "generativelanguage.googleapis.com/generate_content_paid_tier_input_token_count",\n            "quotaId": "GenerateContentPaidTierInputTokensPerModelPerMinute",\n            "quotaDimensions": {\n              "location": "global",\n              "model": "gemma-4-31b"\n            },\n            "quotaValue": "16000"\n          }\n        ]\n      },\n      {\n        "@type": "type.googleapis.com/google.rpc.RetryInfo",\n        "retryDelay": "43s"\n      }\n    ]\n  }\n}\n', 'provider_name': 'Google AI Studio', 'is_byok': True, 'provider_error_code': '429', 'limit_source': 'upstream_provider_account', 'remedy_hint': 'The provider rate-limited your own key. Check the limits on your provider account, or remove the key to use OpenRouter capacity: https://openrouter.ai/settings/integrations', 'previous_errors': [{'code': 429, 'message': 'Rate limit exceeded: free-models-per-day-high-balance. '}]}}, 'user_id': 'user_3IcFCaoAteVd0nzb09NtJTNZ5fM'}. Default move played.
  10. Turn 9
    Gemma 4 31B used Earthquakeautopilot
    Incident report
    [API ERROR] openrouter google/gemma-4-31b-it:free: HTTP 429: Error code: 429 - {'error': {'message': 'Provider returned error', 'code': 429, 'metadata': {'raw': '{\n  "error": {\n    "code": 429,\n    "message": "You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit. \\n* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_paid_tier_input_token_count, limit: 16000, model: gemma-4-31b\\nPlease retry in 43.888052471s.",\n    "status": "RESOURCE_EXHAUSTED",\n    "details": [\n      {\n        "@type": "type.googleapis.com/google.rpc.Help",\n        "links": [\n          {\n            "description": "Learn more about Gemini API quotas",\n            "url": "https://ai.google.dev/gemini-api/docs/rate-limits"\n          }\n        ]\n      },\n      {\n        "@type": "type.googleapis.com/google.rpc.QuotaFailure",\n        "violations": [\n          {\n            "quotaMetric": "generativelanguage.googleapis.com/generate_content_paid_tier_input_token_count",\n            "quotaId": "GenerateContentPaidTierInputTokensPerModelPerMinute",\n            "quotaDimensions": {\n              "location": "global",\n              "model": "gemma-4-31b"\n            },\n            "quotaValue": "16000"\n          }\n        ]\n      },\n      {\n        "@type": "type.googleapis.com/google.rpc.RetryInfo",\n        "retryDelay": "43s"\n      }\n    ]\n  }\n}\n', 'provider_name': 'Google AI Studio', 'is_byok': True, 'provider_error_code': '429', 'limit_source': 'upstream_provider_account', 'remedy_hint': 'The provider rate-limited your own key. Check the limits on your provider account, or remove the key to use OpenRouter capacity: https://openrouter.ai/settings/integrations', 'previous_errors': [{'code': 429, 'message': 'Rate limit exceeded: free-models-per-day-high-balance. '}]}}, 'user_id': 'user_3IcFCaoAteVd0nzb09NtJTNZ5fM'}. Default move played.
    GPT-6 Luna Decisions used Dragonclaw1,535 tok · 0.4 s · $0.0002

    65% MOVE: dragonclaw, 35% MOVE: earthquake, 0% MOVE: stoneedge