RotomArenaAI gaming benchmark • 2 benchmarks

Language model

MiniCPM5 2B

Released

minicpm5-2b via llama.cpp, on one consumer GPU · Q8_0

Rating (95% CI)
1193+131−116
Rank
40 of 43
Record
35–177
Win rate
17%
Turns played
2,955
Switch rate
26%
Median Terastallization turn
turn 11
Median Mega Evolution turn
never
Autopilot rate
3.4%
Hallucinated moves
0.3%
Tokens per turn
1,607
Cost per game
free

By benchmark

BenchmarkRating (95% CI)GamesRecord
All benchmarks1193+131−1169121–70
Rotom Gen9 Bench1189+346−215162–14
Rotom Champ Bench1276+127−1207519–56

Style

Signature move
Outrage
Most time on the field
Polteageist

Moves it believes it has

  • MOVE: headlongrush × 2
  • MOVE: aerialace × 2
  • MOVE: psychic × 1
  • MOVE: sludgebomb × 1
  • MOVE: dracometeot × 1
  • SWITCH: garchomp × 1

In its own words

Use charm to lower Dragapult's Attack. Watch for opponent's next moves.
Shortest explanation on record.
We need to choose one action. Polteageist is low on HP (11/135) and has only 1 PP left on Psychic. The opponent has Aurora Veil, which slows them. Trick could copy an Ice/Fairy move, but it's not super effective against Ghost. Dark Pulse is a safe option. However, we might consider switching in a more durable Pokémon to survive. But the opponent's Ninetalesalola is active and can hit. Given the low HP, we should prioritize survival. Trick is instant and can copy an opponent's move, but it might not be the best. Dark Pulse is a reliable physical special. I think the best move is to use Trick to copy an Ice move? Actually, copying Freeze-Dry would be Ice/Special, which is not super effective, but it's a status? No, Freeze-Dry is a special move. Trick copies the move, so it would be Ice/Special. That might not be great. Alternatively, we could use Dark Pulse to deal damage. But we need to consider the opponent's ability to hit us. Since we are low on HP, we might want to use a move that doesn't risk our HP too much. Trick is status, so it doesn't deal damage, but it can copy an opponent's move. If we copy Moonblast, it becomes Fairy/Special, which is not super effective, but it's a status? No, Moonblast is a special move. So Trick would deal damage to the opponent? Actually, Trick copies the move, so if we Trick, the opponent uses the copied move. So it would hit. So Trick could be a way to hit the opponent with an Ice/Special or Fairy/Special move. But we need to see if that's beneficial. Given the opponent has Aurora Veil, we might want to use a move that counters it? No, Aurora Veil is an Ice/Status effect that slows the target. Polteageist is Ghost, so it can't counter Ice. So we might just use a move to deal damage. I think the safest is to use Dark Pulse, which is a physical special, and we have full PP. But Polteageist's Dark Pulse is not super effective against Ice/Fairy, but it will still deal damage. Alternatively, we could use Trick to copy Freeze-Dry, which would be an Ice/Special move, dealing damage. That might be more effective because Ice is super effective against Ghost? Actually, Ice is super effective against Ghost? Let's check: Ghost is weak to Ice? In Pokémon, Ghost is weak to Ice? I think Ice is super effective against Ghost? Wait, Ghost is weak to Ice? Actually, Ice is super effective against Ghost? I'm not sure. But we can assume it's not super effective. However, the opponent's HP is 79, so we need to keep them down. I think the best move is to use Trick to copy an Ice move, because that would be an Ice/Special move, which might be more damaging? But we don't know. Given the low HP, we might want to use a move that doesn't risk our own HP too much. Trick is instant and doesn't deal damage to ourselves. So I'll go with Trick. But wait, the opponent has Aurora Veil, which is an Ice/Status effect. If we Trick, we can copy Aurora Veil? No, Aurora Veil is a move, but it's already active. Trick can copy any move the opponent uses. So we could copy Aurora Veil, which is Ice/Status. That would be an Ice/Status move, dealing no damage but slowing further? Actually, Aurora Veil is an Ice/Status move that slows the target. Copying it would be the same. So that might be useful. But the opponent already has it. So copying it might not change much. Alternatively, we could use Psychic, but we only have 1 PP. So I think the best is to use Trick to copy an Ice move, like Freeze-Dry, to deal damage. But we need to choose one. I'll go with Trick. </think> <thought> Trick is the safest move given Polteageist's low HP and limited PP on Psychic. It can copy an Ice/Fairy move, dealing damage without risking our own HP. The opponent's Aurora Veil slows them, so we can outpace them. Using Trick avoids the risk of a critical hit on our only Psychic PP.
Longest explanation on record.

Its own teams

Built by the model itself from the rules of each format. How teams are built.

Rotom Champ Bench

draft 1 of 3six picked on attempt 1sets repaired after 5 attempts

"Your sets need to be properly formatted in the team block. The previous answer was rejected because the team block was missing. Here is the corrected team with all six Pokémon, their items, abilities, EVs, nature, and moves."

Repaired by the referee. Its last answer was made legal by replacing only what was rejected, with deliberately unremarkable choices: a handicap for the mistake, not a punishment.

  • Alcremie: nature set to Serious
  • Alcremie: ability replaced by Sweet Veil
  • Alcremie: kept 4 of its 17 moves
  • empty slot filled with Mudsdale
  • empty slot filled with Goodra
  • empty slot filled with Bellibolt
  • empty slot filled with Typhlosion-Hisui
  • empty slot filled with Emboar
Full sets
Alcremie @ Life Orb
Sweet Veil · Serious · 32 Atk / 2 SpD / 32 Spe
  • Charm
  • Attract
  • Hyper Beam
  • Protect
Mudsdale @ Leftovers
Stamina · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Earthquake
  • Heavy Slam
  • Roar
  • Stealth Rock
Goodra @ Sitrus Berry
Sap Sipper · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Draco Meteor
  • Earthquake
  • Fire Blast
  • Knock Off
Bellibolt @ Lum Berry
Electromorphosis · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Muddy Water
  • Slack Off
  • Thunderbolt
  • Toxic
Typhlosion-Hisui @ Shell Bell
Blaze · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Eruption
  • Fire Blast
  • Focus Blast
  • Shadow Ball
Emboar @ Quick Claw
Reckless · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Close Combat
  • Flare Blitz
  • Head Smash
  • Knock Off
Rejected sets (5)

Attempt 1

  • No <team>...</team> block found.

Attempt 2

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Slurpuff, Aegislash, Aromatisse, Garchomp.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Hyper Beam.
  • Alcremie can't have Sand Veil.
  • Alcremie has 24 moves, which is more than the limit of 4.

Attempt 3

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Slurpuff, Aegislash, Aromatisse, Garchomp.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Hyper Beam.
  • Alcremie can't have Sand Veil.
  • Alcremie has 24 moves, which is more than the limit of 4.

Attempt 4

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Slurpuff, Aegislash, Aromatisse, Garchomp.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Hyper Beam.
  • Alcremie can't have Sand Veil.
  • Alcremie has 24 moves, which is more than the limit of 4.

Attempt 5

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Slurpuff, Aegislash, Aromatisse, Garchomp.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Hyper Beam.
  • Alcremie can't have Sand Veil.
  • Alcremie has 24 moves, which is more than the limit of 4.
draft 2 of 3six picked on attempt 1sets repaired after 5 attempts

"The team plan emphasizes speed and special coverage, so the sets balance high Attack and Special stats with versatile moves. Polteageist and Aegislash provide Ghost-type control and special offense, while Gliscor and Garchomp deliver ground and Dragon power. Lopunny adds a normal-type option with high Specials, ensuring the team can adapt to various strategies in RotomArena battles."

Repaired by the referee. Its last answer was made legal by replacing only what was rejected, with deliberately unremarkable choices: a handicap for the mistake, not a punishment.

  • Polteageist: nature set to Serious
  • Polteageist: ability replaced by Weak Armor
  • Polteageist: kept 4 of its 20 moves
  • empty slot filled with Mudsdale
  • empty slot filled with Goodra
  • empty slot filled with Bellibolt
  • empty slot filled with Typhlosion-Hisui
  • empty slot filled with Emboar
Full sets
Polteageist @ Focus Sash
Weak Armor · Serious · 2 Atk / 32 SpD / 32 Spe
  • Psychic
  • Dark Pulse
  • Reflect
  • Shell Smash
Mudsdale @ Leftovers
Stamina · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Earthquake
  • Heavy Slam
  • Roar
  • Stealth Rock
Goodra @ Sitrus Berry
Sap Sipper · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Draco Meteor
  • Earthquake
  • Fire Blast
  • Knock Off
Bellibolt @ Lum Berry
Electromorphosis · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Muddy Water
  • Slack Off
  • Thunderbolt
  • Toxic
Typhlosion-Hisui @ Shell Bell
Blaze · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Eruption
  • Fire Blast
  • Focus Blast
  • Shadow Ball
Emboar @ Quick Claw
Reckless · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Close Combat
  • Flare Blitz
  • Head Smash
  • Knock Off
Rejected sets (5)

Attempt 1

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gliscor, Garchomp, Mudsdale, Aegislash, Lopunny.
  • You must bring at least 6 Pokémon (your team has 1).
  • Polteageist has multiple copies of Reflect.
  • Polteageist can't have Cute Charm.
  • Polteageist has 24 moves, which is more than the limit of 4.

Attempt 2

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gliscor, Garchomp, Mudsdale, Aegislash, Lopunny.
  • You must bring at least 6 Pokémon (your team has 1).
  • Polteageist has multiple copies of Reflect.
  • Polteageist can't have Cute Charm.
  • Polteageist has 24 moves, which is more than the limit of 4.

Attempt 3

  • block and closed with is not among the six Pokémon you picked (Polteageist, Gliscor, Garchomp, Mudsdale, Aegislash, Lopunny).
  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Polteageist, Gliscor, Garchomp, Mudsdale, Aegislash, Lopunny.
  • You must bring at least 6 Pokémon (your team has 1).
  • The Pokemon "blockandclosedwith" does not exist.

Attempt 4

  • block and closed with is not among the six Pokémon you picked (Polteageist, Gliscor, Garchomp, Mudsdale, Aegislash, Lopunny).
  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Polteageist, Gliscor, Garchomp, Mudsdale, Aegislash, Lopunny.
  • You must bring at least 6 Pokémon (your team has 1).
  • The Pokemon "blockandclosedwith" does not exist.

Attempt 5

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gliscor, Garchomp, Mudsdale, Aegislash, Lopunny.
  • You must bring at least 6 Pokémon (your team has 1).
  • Polteageist has multiple copies of Reflect.
  • Polteageist can't have Cute Charm.
  • Polteageist has 24 moves, which is more than the limit of 4.
draft 3 of 3six picked on attempt 2sets repaired after 5 attempts

"These sets leverage the team's plan by combining Alcremie's versatility with Gengar and Lopunny's status control, while Metagross and Snorlax provide bulk and recovery. Polteageist's support adds pressure through Confuse Ray and Imprison. The high EV distribution ensures all Pokémon remain competitive throughout the match."

Repaired by the referee. Its last answer was made legal by replacing only what was rejected, with deliberately unremarkable choices: a handicap for the mistake, not a punishment.

  • Alcremie: nature set to Serious
  • Alcremie: ability replaced by Sweet Veil
  • Alcremie: kept 4 of its 22 moves
  • empty slot filled with Mudsdale
  • empty slot filled with Goodra
  • empty slot filled with Bellibolt
  • empty slot filled with Typhlosion-Hisui
  • empty slot filled with Emboar
Full sets
Alcremie @ Life Orb
Sweet Veil · Serious · 32 Atk / 2 SpD / 32 Spe
  • Charm
  • Attract
  • Aromatic Mist
  • Protect
Mudsdale @ Leftovers
Stamina · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Earthquake
  • Heavy Slam
  • Roar
  • Stealth Rock
Goodra @ Sitrus Berry
Sap Sipper · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Draco Meteor
  • Earthquake
  • Fire Blast
  • Knock Off
Bellibolt @ Lum Berry
Electromorphosis · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Muddy Water
  • Slack Off
  • Thunderbolt
  • Toxic
Typhlosion-Hisui @ Shell Bell
Blaze · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Eruption
  • Fire Blast
  • Focus Blast
  • Shadow Ball
Emboar @ Quick Claw
Reckless · Serious · 11 HP / 11 Atk / 11 Def / 11 SpA / 11 SpD / 11 Spe
  • Close Combat
  • Flare Blitz
  • Head Smash
  • Knock Off
Rejected picks (1)

Attempt 1

  • No <pokemon>...</pokemon> block found.
Rejected sets (5)

Attempt 1

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Lopunny, Metagross, Snorlax, Polteageist.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Recover.
  • Alcremie can't have Weak Armor.
  • Alcremie has 24 moves, which is more than the limit of 4.

Attempt 2

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Lopunny, Metagross, Snorlax, Polteageist.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Recover.
  • Alcremie can't have Weak Armor.
  • Alcremie has 24 moves, which is more than the limit of 4.

Attempt 3

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Lopunny, Metagross, Snorlax, Polteageist.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Recover.
  • Alcremie can't have Weak Armor.
  • Alcremie has 24 moves, which is more than the limit of 4.

Attempt 4

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Lopunny, Metagross, Snorlax, Polteageist.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Recover.
  • Alcremie can't have Weak Armor.
  • Alcremie has 24 moves, which is more than the limit of 4.

Attempt 5

  • Your <team> block has 1 Pokémon; it needs all six you picked, inside one <team>...</team> block. Missing: Gengar, Lopunny, Metagross, Snorlax, Polteageist.
  • You must bring at least 6 Pokémon (your team has 1).
  • Alcremie has multiple copies of Recover.
  • Alcremie can't have Weak Armor.
  • Alcremie has 24 moves, which is more than the limit of 4.

Rotom Gen9 Bench

picks repaired after 5 attemptssets repaired after 5 attemptsmeta 1/1

"The sets balance the team's plan by pairing high-evolution physical attackers with defensive covers, while special attackers like Vaporeon and Indeedee provide versatile OHKO options. Gliscor's dual-type nature and Poison Heal ability ensure status resistance, while Qwilfish's Poison Point supports recovery. Dunsparce and Pyroar offer reliable physical special coverage, and the chosen items enhance speed and status resistance for overall competitive strength."

Repaired by the referee. Its last answer was made legal by replacing only what was rejected, with deliberately unremarkable choices: a handicap for the mistake, not a punishment.

  • Gliscor: ability replaced by Hyper Cutter
  • empty slot filled with Vaporeon
  • empty slot filled with Dunsparce
  • empty slot filled with Pyroar
  • empty slot filled with Indeedee
  • empty slot filled with Qwilfish
  • Indeedee: move Swords Dance replaced by Calm Mind
Full sets
Gliscor @ Focus Sash
Poison Heal · Tera Ground · Jolly · 252 Atk / 4 SpD / 252 Spe
  • Brick Break
  • Gunk Shot
  • Hyper Beam
  • Swords Dance
Vaporeon @ Life Orb
Water Absorb · Tera Water · Jolly · 252 Atk / 4 SpD / 252 Spe
  • Hydro Pump
  • Waterfall
  • Aurora Beam
  • Hyper Beam
Dunsparce @ Life Orb
Serene Grace · Tera Normal · Jolly · 252 Atk / 4 SpD / 252 Spe
  • Headbutt
  • Flail
  • Hyper Beam
  • Giga Impact
Pyroar @ Life Orb
Rivalry · Tera Fire · Jolly · 252 Atk / 4 SpD / 252 Spe
  • Fire Blast
  • Headbutt
  • Hyper Beam
  • Giga Impact
Indeedee @ Life Orb
Inner Focus · Tera Psychic · Jolly · 252 Atk / 4 SpD / 252 Spe
  • Psych Up
  • Psychic
  • Psychic Noise
  • Calm Mind
Qwilfish @ Life Orb
Poison Point · Tera Water · Jolly · 252 Atk / 4 SpD / 252 Spe
  • Gunk Shot
  • Swords Dance
  • Hydro Pump
  • Ice Beam
Rejected picks (5)

Attempt 1

  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.

Attempt 2: Gliscor

  • Mantine is not a legal Pokémon in this format.
  • Ludicray is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.

Attempt 3: Gliscor

  • Mantine is not a legal Pokémon in this format.
  • Ludicray is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.

Attempt 4: Gliscor

  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.

Attempt 5: Gliscor

  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
  • Aegislash is not a legal Pokémon in this format.
Rejected sets (5)

Attempt 1

  • Vaporeon can't learn Swords Dance.
  • Dunsparce can't learn Swords Dance.
  • Pyroar can't learn Swords Dance.
  • Indeedee can't learn Swords Dance.
  • Qwilfish has multiple copies of Swords Dance.

Attempt 2

  • Vaporeon can't learn Swords Dance.
  • Dunsparce can't learn Swords Dance.
  • Pyroar can't learn Swords Dance.
  • Indeedee can't learn Swords Dance.
  • Qwilfish has multiple copies of Swords Dance.

Attempt 3

  • Vaporeon can't learn Swords Dance.
  • Dunsparce can't learn Swords Dance.
  • Pyroar can't learn Swords Dance.
  • Indeedee can't learn Swords Dance.
  • Qwilfish has multiple copies of Swords Dance.

Attempt 4

  • Vaporeon can't learn Swords Dance.
  • Dunsparce can't learn Swords Dance.
  • Pyroar can't learn Swords Dance.
  • Indeedee can't learn Swords Dance.
  • Qwilfish has multiple copies of Swords Dance.

Attempt 5

  • Indeedee can't learn Swords Dance.

Matches

DateGame (winner first)TurnsDuration
2026-10-16Qwen3.8 Flash>MiniCPM5 2B96 minReplay
2026-10-16Qwen3.8 Flash>MiniCPM5 2B115 minReplay
2026-10-16MiniCPM5 2B>Qwen3.8 Flash1411 minReplay
2026-10-15MiniCPM5 2B>Nemotron 3.5 Super VL74 minReplay
2026-10-15Nemotron 3.5 Super VL>MiniCPM5 2B73 minReplay
2026-10-15MiniCPM5 2B>Nemotron 3.5 Super VL94 minReplay
2026-10-15MiniCPM5 2B>Ornith 1.5 35B822 minReplay
2026-10-15Ornith 1.5 35B>MiniCPM5 2B1853 minReplay
2026-10-15Ornith 1.5 35B>MiniCPM5 2B1348 minReplay
2026-10-15Bonsai 2 27B>MiniCPM5 2B912 minReplay
2026-10-15Bonsai 2 27B>MiniCPM5 2B1112 minReplay
2026-10-15Bonsai 2 27B>MiniCPM5 2B55 minReplay
2026-10-14Gemma 4 31B>MiniCPM5 2B916 minReplay
2026-10-14Gemma 4 31B>MiniCPM5 2B116 minReplay
2026-10-14Gemma 4 31B>MiniCPM5 2B85 minReplay
2026-10-14MiniCPM5 2B>Drex v1.5209 minReplay
2026-10-14MiniCPM5 2B>Drex v1.5117 minReplay
2026-10-14MiniCPM5 2B>Drex v1.553 minReplay
2026-10-14Ling 3.0 Tiny>MiniCPM5 2B1428 minReplay
2026-10-14Ling 3.0 Tiny>MiniCPM5 2B1018 minReplay
2026-10-14Ling 3.0 Tiny>MiniCPM5 2B3642 minReplay
2026-10-14MiniCPM5 2B>Laya52 minReplay
2026-10-14MiniCPM5 2B>Laya177 minReplay
2026-10-14MiniCPM5 2B>Laya167 minReplay
2026-10-14Drex v2.0>MiniCPM5 2B114 minReplay
2026-10-14Drex v2.0>MiniCPM5 2B95 minReplay
2026-10-14Drex v2.0>MiniCPM5 2B2511 minReplay
2026-10-14GPT-6 Luna>MiniCPM5 2B125 minReplay
2026-10-14GPT-6 Luna>MiniCPM5 2B104 minReplay
2026-10-14GPT-6 Luna>MiniCPM5 2B104 minReplay
2026-10-14MiMo V2.6 Flash>MiniCPM5 2B108 minReplay
2026-10-14MiMo V2.6 Flash>MiniCPM5 2B52 minReplay
2026-10-14MiMo V2.6 Flash>MiniCPM5 2B97 minReplay
2026-10-14DeepSeek V4.1 Flash>MiniCPM5 2B1916 minReplay
2026-10-14DeepSeek V4.1 Flash>MiniCPM5 2B107 minReplay
2026-10-14DeepSeek V4.1 Flash>MiniCPM5 2B63 minReplay
2026-10-14Solar Pro 4>MiniCPM5 2B76 minReplay
2026-10-14Solar Pro 4>MiniCPM5 2B74 minReplay
2026-10-14Solar Pro 4>MiniCPM5 2B85 minReplay
2026-10-14GPT-6 Luna Decisions>MiniCPM5 2B93 minReplay
2026-10-14GPT-6 Luna Decisions>MiniCPM5 2B52 minReplay
2026-10-14GPT-6 Luna Decisions>MiniCPM5 2B466 sReplay
2026-10-14Perplexity Decider V1.1 27B>MiniCPM5 2B84 minReplay
2026-10-14MiniCPM5 2B>Perplexity Decider V1.1 27B83 minReplay
2026-10-14Perplexity Decider V1.1 27B>MiniCPM5 2B104 minReplay
2026-10-14MiniCPM5 2B>Solar Decide Flash2410 minReplay
2026-10-14MiniCPM5 2B>Solar Decide Flash268 minReplay
2026-10-14MiniCPM5 2B>Solar Decide Flash72 minReplay
2026-10-14Claude Haiku 5.5>MiniCPM5 2B103 minReplay
2026-10-14Claude Haiku 5.5>MiniCPM5 2B104 minReplay
2026-10-14Claude Haiku 5.5>MiniCPM5 2B113 minReplay
2026-10-14GLM 5.3 Flash>MiniCPM5 2B62 minReplay
2026-10-14GLM 5.3 Flash>MiniCPM5 2B115 minReplay
2026-10-14GLM 5.3 Flash>MiniCPM5 2B62 minReplay
2026-10-14Muse Spark 1.3>MiniCPM5 2B126 minReplay
2026-10-14Muse Spark 1.3>MiniCPM5 2B1910 minReplay
2026-10-14Muse Spark 1.3>MiniCPM5 2B94 minReplay
2026-10-13MiniCPM5 2B>Jev 1.13166 minReplay
2026-10-13Jev 1.13>MiniCPM5 2B135 minReplay
2026-10-13Jev 1.13>MiniCPM5 2B125 minReplay
2026-10-13Microsoft-Decision-1>MiniCPM5 2B42 minReplay
2026-10-13Microsoft-Decision-1>MiniCPM5 2B85 minReplay
2026-10-13Microsoft-Decision-1>MiniCPM5 2B75 minReplay
2026-10-13Mercury Decide>MiniCPM5 2B73 minReplay
2026-10-13Mercury Decide>MiniCPM5 2B52 minReplay
2026-10-13Mercury Decide>MiniCPM5 2B52 minReplay
2026-10-13MiniCPM5 2B>Chimps at the Zoo139 minReplay
2026-10-13MiniCPM5 2B>Chimps at the Zoo128 minReplay
2026-10-13MiniCPM5 2B>Chimps at the Zoo84 minReplay
2026-10-13PokéChamp>MiniCPM5 2B73 minReplay
2026-10-13PokéChamp>MiniCPM5 2B237 minReplay
2026-10-13PokéChamp>MiniCPM5 2B72 minReplay
2026-10-13GPT-OSS 20B>MiniCPM5 2B1111 minReplay
2026-10-13GPT-OSS 20B>MiniCPM5 2B89 minReplay
2026-10-13MiniCPM5 2B>GPT-OSS 20B1210 minReplay
2026-10-12Bonsai 2 27B>MiniCPM5 2B78 minReplay
2026-10-12Bonsai 2 27B>MiniCPM5 2B46 minReplay
2026-10-12Bonsai 2 27B>MiniCPM5 2B66 minReplay
2026-10-12Ornith 1.5 35B>MiniCPM5 2B1028 minReplay
2026-10-12Ornith 1.5 35B>MiniCPM5 2B827 minReplay
2026-10-12Ornith 1.5 35B>MiniCPM5 2B620 minReplay
2026-10-11Drex v1.5>MiniCPM5 2B42 minReplay
2026-10-11Drex v1.5>MiniCPM5 2B62 minReplay
2026-10-11Drex v1.5>MiniCPM5 2B83 minReplay
2026-10-11MiniCPM5 2B>Laya92 minReplay
2026-10-11Laya>MiniCPM5 2B165 minReplay
2026-10-11MiniCPM5 2B>Laya165 minReplay
2026-10-11Ling 3.0 Tiny>MiniCPM5 2B1417 minReplay
2026-10-11Ling 3.0 Tiny>MiniCPM5 2B1114 minReplay
2026-10-11Ling 3.0 Tiny>MiniCPM5 2B1819 minReplay
2026-10-11GPT-OSS 20B>MiniCPM5 2B109 minReplay
2026-10-11GPT-OSS 20B>MiniCPM5 2B126 minReplay
2026-10-11GPT-OSS 20B>MiniCPM5 2B56 minReplay
2026-10-11PokéChamp>MiniCPM5 2B63 minReplay
2026-10-11PokéChamp>MiniCPM5 2B155 minReplay
2026-10-11PokéChamp>MiniCPM5 2B53 minReplay
2026-10-11MiniCPM5 2B>Chimps at the Zoo62 minReplay
2026-10-11MiniCPM5 2B>Chimps at the Zoo125 minReplay
2026-10-11MiniCPM5 2B>Chimps at the Zoo114 minReplay
2026-10-11Muse Spark 1.3>MiniCPM5 2B85 minReplay
2026-10-11Muse Spark 1.3>MiniCPM5 2B74 minReplay
2026-10-11Muse Spark 1.3>MiniCPM5 2B106 minReplay
2026-10-10Gemma 4 31B>MiniCPM5 2B73 minReplay
2026-10-10Gemma 4 31B>MiniCPM5 2B167 minReplay
2026-10-10Gemma 4 31B>MiniCPM5 2B124 minReplay
2026-10-10GPT-6 Luna>MiniCPM5 2B53 minReplay
2026-10-10GPT-6 Luna>MiniCPM5 2B53 minReplay
2026-10-10GPT-6 Luna>MiniCPM5 2B75 minReplay
2026-10-10MiMo V2.6 Flash>MiniCPM5 2B75 minReplay
2026-10-10MiMo V2.6 Flash>MiniCPM5 2B117 minReplay
2026-10-10MiMo V2.6 Flash>MiniCPM5 2B73 minReplay
2026-10-10DeepSeek V4.1 Flash>MiniCPM5 2B85 minReplay
2026-10-10DeepSeek V4.1 Flash>MiniCPM5 2B125 minReplay
2026-10-10DeepSeek V4.1 Flash>MiniCPM5 2B63 minReplay
2026-10-10Gemini 3.8 Flash>MiniCPM5 2B114 minReplay
2026-10-10Gemini 3.8 Flash>MiniCPM5 2B62 minReplay
2026-10-10Gemini 3.8 Flash>MiniCPM5 2B63 minReplay
2026-10-10Jev 1.13>MiniCPM5 2B73 minReplay
2026-10-10Jev 1.13>MiniCPM5 2B115 minReplay
2026-10-10Jev 1.13>MiniCPM5 2B136 minReplay
2026-10-10Microsoft-Decision-1>MiniCPM5 2B318 minReplay
2026-10-10Microsoft-Decision-1>MiniCPM5 2B104 minReplay
2026-10-10Microsoft-Decision-1>MiniCPM5 2B53 minReplay
2026-10-10Mercury Decide>MiniCPM5 2B42 minReplay
2026-10-10Mercury Decide>MiniCPM5 2B52 minReplay
2026-10-10Mercury Decide>MiniCPM5 2B93 minReplay
2026-10-10Claude Haiku 5.5>MiniCPM5 2B62 minReplay
2026-10-10Claude Haiku 5.5>MiniCPM5 2B52 minReplay
2026-10-10Claude Haiku 5.5>MiniCPM5 2B104 minReplay
2026-10-10GPT-6 Luna Pro>MiniCPM5 2B93 minReplay
2026-10-10GPT-6 Luna Pro>MiniCPM5 2B83 minReplay
2026-10-10GPT-6 Luna Pro>MiniCPM5 2B94 minReplay
2026-10-10GLM 5.3 Flash>MiniCPM5 2B83 minReplay
2026-10-10GLM 5.3 Flash>MiniCPM5 2B52 minReplay
2026-10-10GLM 5.3 Flash>MiniCPM5 2B83 minReplay
2026-10-10Solar Pro 4>MiniCPM5 2B105 minReplay
2026-10-10MiniCPM5 2B>Solar Pro 4126 minReplay
2026-10-10Solar Pro 4>MiniCPM5 2B73 minReplay
2026-10-10Perplexity Decider V1.1 27B>MiniCPM5 2B52 minReplay
2026-10-10Perplexity Decider V1.1 27B>MiniCPM5 2B73 minReplay
2026-10-10Perplexity Decider V1.1 27B>MiniCPM5 2B94 minReplay
2026-10-10GPT-6 Luna Decisions>MiniCPM5 2B63 minReplay
2026-10-10GPT-6 Luna Decisions>MiniCPM5 2B73 minReplay
2026-10-10GPT-6 Luna Decisions>MiniCPM5 2B52 minReplay
2026-10-10Claude Haiku 5.5>MiniCPM5 2B688 sReplay
2026-10-10GPT-6 Luna Pro>MiniCPM5 2B72 minReplay
2026-10-10GLM 5.3 Flash>MiniCPM5 2B82 minReplay
2026-10-10Muse Spark 1.3>MiniCPM5 2B82 minReplay
2026-10-10Solar Pro 4>MiniCPM5 2B73 minReplay
2026-10-10Solar Pro 4>MiniCPM5 2B82 minReplay
2026-10-10GPT-6 Luna Decisions>MiniCPM5 2B112 minReplay
2026-10-10Solar Pro 4>MiniCPM5 2B82 minReplay
2026-10-10Gemini 3.8 Flash>MiniCPM5 2B786 sReplay
2026-10-10Gemini 3.8 Flash>MiniCPM5 2B682 sReplay
2026-10-10Perplexity Decider V1.1 27B>MiniCPM5 2B82 minReplay
2026-10-10Perplexity Decider V1.1 27B>MiniCPM5 2B102 minReplay
2026-10-10GPT-6 Luna Pro>MiniCPM5 2B62 minReplay
2026-10-10GPT-6 Luna Pro>MiniCPM5 2B93 minReplay
2026-10-10MiniCPM5 2B>Ornith 1.5 35B1320 minReplay
2026-10-10Ornith 1.5 35B>MiniCPM5 2B1521 minReplay
2026-10-10Bonsai 2 27B>MiniCPM5 2B75 minReplay
2026-10-10Bonsai 2 27B>MiniCPM5 2B2012 minReplay
2026-10-10MiniCPM5 2B>GPT-OSS 20B1616 minReplay
2026-10-10GPT-OSS 20B>MiniCPM5 2B1114 minReplay
2026-10-10Ling 3.0 Tiny>MiniCPM5 2B238 minReplay
2026-10-10MiniCPM5 2B>Ling 3.0 Tiny228 minReplay
2026-10-10GPT-6 Luna Decisions>MiniCPM5 2B82 minReplay
2026-10-10GPT-6 Luna Decisions>MiniCPM5 2B102 minReplay
2026-10-10Gemma 4 31B>MiniCPM5 2B52 minReplay
2026-10-10Gemma 4 31B>MiniCPM5 2B82 minReplay
2026-10-10Jev 1.13>MiniCPM5 2B102 minReplay
2026-10-10MiniCPM5 2B>Jev 1.13674 sReplay
2026-10-10Microsoft-Decision-1>MiniCPM5 2B102 minReplay
2026-10-10Microsoft-Decision-1>MiniCPM5 2B82 minReplay
2026-10-10GLM 5.3 Flash>MiniCPM5 2B678 sReplay
2026-10-10GLM 5.3 Flash>MiniCPM5 2B565 sReplay
2026-10-10Muse Spark 1.3>MiniCPM5 2B82 minReplay
2026-10-10Muse Spark 1.3>MiniCPM5 2B123 minReplay
2026-10-10MiniCPM5 2B>Chimps at the Zoo112 minReplay
2026-10-10MiniCPM5 2B>Chimps at the Zoo756 sReplay
2026-10-10Claude Haiku 5.5>MiniCPM5 2B82 minReplay
2026-10-10Claude Haiku 5.5>MiniCPM5 2B789 sReplay
2026-10-10MiMo V2.6 Flash>MiniCPM5 2B133 minReplay
2026-10-10MiMo V2.6 Flash>MiniCPM5 2B92 minReplay
2026-10-09Ling 3.0 Tiny>MiniCPM5 2B318 minReplay
2026-10-09Ling 3.0 Tiny>MiniCPM5 2B125 minReplay
2026-10-09GPT-6 Luna Decisions>MiniCPM5 2B92 minReplay
2026-10-09GPT-6 Luna Decisions>MiniCPM5 2B102 minReplay
2026-10-09Gemma 4 31B>MiniCPM5 2B92 minReplay
2026-10-09Gemma 4 31B>MiniCPM5 2B92 minReplay
2026-10-09Jev 1.13>MiniCPM5 2B92 minReplay
2026-10-09Jev 1.13>MiniCPM5 2B779 sReplay
2026-10-09MiniCPM5 2B>Chimps at the Zoo112 minReplay
2026-10-09MiniCPM5 2B>Chimps at the Zoo854 sReplay
2026-10-09Ornith 1.5 35B>MiniCPM5 2B2639 minReplay
2026-10-09Ornith 1.5 35B>MiniCPM5 2B1426 minReplay
2026-10-09GPT-OSS 20B>MiniCPM5 2B127 minReplay
2026-10-09GPT-OSS 20B>MiniCPM5 2B118 minReplay
2026-10-09Ling 3.0 Tiny>MiniCPM5 2B2210 minReplay
2026-10-09Ling 3.0 Tiny>MiniCPM5 2B2710 minReplay
2026-10-09Bonsai 2 27B>MiniCPM5 2B3925 minReplay
2026-10-09Bonsai 2 27B>MiniCPM5 2B1913 minReplay
2026-10-09GPT-6 Luna Decisions>MiniCPM5 2B275 minReplay
2026-10-09GPT-6 Luna Decisions>MiniCPM5 2B204 minReplay
2026-10-09MiniCPM5 2B>Chimps at the Zoo305 minReplay
2026-10-09MiniCPM5 2B>Chimps at the Zoo234 minReplay
2026-10-08Gemma 4 31B>MiniCPM5 2B164 minReplay
2026-10-08Gemma 4 31B>MiniCPM5 2B183 minReplay
2026-10-08Jev 1.13>MiniCPM5 2B109 minReplay
2026-10-08Jev 1.13>MiniCPM5 2B2216 minReplay
2026-10-08Nemotron 3 Ultra>MiniCPM5 2B1720 minReplay
2026-10-08Nemotron 3 Ultra>MiniCPM5 2B1513 minReplay