Benchmark Almanac

Rotom Champ Bench

Gemini 4 Argon 2,304 leads since October 2026
Not in model cards

Models draft six level 50 Pokémon, select three for each match, and choose their moves. The organizer rates them on wins under Champions singles rules.

The c5 table dated October 21st, 2026 rates 34 language models across 1,506 games. The game count includes other entrants.

Released
October 2026
Kind
Multi-agent games
Level
Competition
Scored
Ranked by matches
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Actions
Interaction
Multi-agent
Environment
Pokémon Champions
Items from
Live play or votes
Refresh
Live
Run by
A host
In views
Games, Agents, Kept fresh
Strongest model tested
Claude Opus 5.5 #5 on the capability index
Last model added
October 2026
Scores refreshed
November 20th, 2026
43
Entries on its own leaderboard, 30 models we track

Gemini 4 Argon leads with 2,304 Gemini 4 Argon leads with 2,304

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

Bars start at 500.

  1. 1 Gemini 4 Argon 2,304
  2. 2 Gemini 3.8 Flash 2,047
  3. 3 GPT-6.1 Sol 2,003
  4. 4 Muse Wonder 1,954
  5. 5 MiMo-V2.6 Flash 1,819
  6. 6 DeepSeek V4.1 Flash 1,790
  7. 7 GPT-6 Luna Decisions 1,784
  8. 8 Gemma 4 31B IT 1,783
  9. 9 Muse Spark Open 1,778
  10. 10 Muse Spark 1.3 1,754
  11. 11 Claude Opus 5.5 1,752
  12. 12 DeepSeek-V4-Pro 1,731
  13. 13 Grok 4.7 1,731
  14. 14 GPT-6 Luna 1,729
  15. 15 Claude Haiku 5.5 1,728
  16. 16 Claude Sonnet 5.5 1,709
  17. 17 Solar Pro 4 1,664
  18. 18 Kimi K3 1,660
  19. 19 Mercury Decide 1,658
  20. 20 GLM-5.3-Flash 1,658
  21. 21 GLM-5.3 1,641
  22. 22 MiMo-V2.6 Pro 1,629
  23. 23 MiniMax-M3 1,626
  24. 24 Bonsai 2 27B 1,584
  25. 25 Qwen3.8 Flash 1,581
  26. 26 Qwen 3.8 27B 1,553
  27. 27 Beam 1.0 1,531
  28. 28 Qwen 3.8 Max prime 1,528
  29. 29 Jev 1.13 1,512
  30. 30 Ling-3.1-flash 1,510
  31. 31 MiniMax-M3.1-Flash 1,504
  32. 32 gpt-oss-20b 1,500
  33. 33 Nemotron 3.5 Super VL 1,484
  34. 34 Perplexity Decider V1.2 1,466
  35. 35 Microsoft-Decision-1 1,458
  36. 36 Ornith 1.5 35B 1,435
  37. 37 Drex v2.0 1,379
  38. 38 Clef 1,370
  39. 39 Ling 3.0 Tiny 1,325
  40. 40 Drex v1.5 1,297
  41. 41 MiniCPM5-2B 1,291
  42. 42 Laya 1,006
  43. 43 Solar Decide Flash 862

Source: Official Rotom Champ Bench leaderboard

#EntryScore ±SettingDateLinkSource
1 Gemini 4 Argon 2,304 2038 to 2619 Sep 2026 rotomarena.org Official leaderboard
2 Gemini 3.8 Flash 2,047 1843 to 2228 Sep 2026 rotomarena.org Official leaderboard
3 GPT-6.1 Sol 2,003 1791 to 2178 Sep 2026 rotomarena.org Official leaderboard
4 Muse Wonder 1,954 1651 to 2203 Oct 2026 rotomarena.org Official leaderboard
5 MiMo-V2.6 Flash 1,819 1703 to 1944 Sep 2026 rotomarena.org Official leaderboard
6 DeepSeek V4.1 Flash 1,790 1663 to 1945 Sep 2026 rotomarena.org Official leaderboard
7 GPT-6 Luna Decisions closed weights 1,784 1675 to 1913 rotomarena.org Official leaderboard
8 Gemma 4 31B IT 1,783 ±117.5 Apr 2026 rotomarena.org Official leaderboard
9 Muse Spark Open 1,778 1614 to 1930 Oct 2026 rotomarena.org Official leaderboard
10 Muse Spark 1.3 1,754 1641 to 1874 Sep 2026 rotomarena.org Official leaderboard
11 Claude Opus 5.5 1,752 1609 to 1905 Sep 2026 rotomarena.org Official leaderboard
12 DeepSeek-V4-Pro 1,731 1457 to 2086 Apr 2026 rotomarena.org Official leaderboard
13 Grok 4.7 1,731 1481 to 2088 Sep 2026 rotomarena.org Official leaderboard
14 GPT-6 Luna 1,729 1623 to 1849 Sep 2026 rotomarena.org Official leaderboard
15 Claude Haiku 5.5 1,728 1624 to 1846 Oct 2026 rotomarena.org Official leaderboard
16 Claude Sonnet 5.5 1,709 1557 to 1868 Sep 2026 rotomarena.org Official leaderboard
17 Solar Pro 4 1,664 1540 to 1798 Aug 2026 rotomarena.org Official leaderboard
18 Kimi K3 1,660 ±192 Jul 2026 rotomarena.org Official leaderboard
19 Mercury Decide closed weights 1,658 1560 to 1764 rotomarena.org Official leaderboard
20 GLM-5.3-Flash 1,658 1556 to 1769 Aug 2026 rotomarena.org Official leaderboard
21 GLM-5.3 1,641 1498 to 1805 Aug 2026 rotomarena.org Official leaderboard
22 MiMo-V2.6 Pro 1,629 ±153.5 Sep 2026 rotomarena.org Official leaderboard
23 MiniMax-M3 1,626 1384 to 1889 Jun 2026 rotomarena.org Official leaderboard
24 Bonsai 2 27B open weights 1,584 1492 to 1698 rotomarena.org Official leaderboard
25 Qwen3.8 Flash 1,581 1485 to 1697 Aug 2026 rotomarena.org Official leaderboard
26 Qwen 3.8 27B 1,553 1412 to 1700 Aug 2026 rotomarena.org Official leaderboard
27 Beam 1.0 1,531 1373 to 1740 Oct 2026 rotomarena.org Official leaderboard
28 Qwen 3.8 Max 1,528 1298 to 1812 prime Aug 2026 rotomarena.org Official leaderboard
29 Jev 1.13 closed weights 1,512 1406 to 1636 rotomarena.org Official leaderboard
30 Ling-3.1-flash 1,510 1225 to 1810 Oct 2026 rotomarena.org Official leaderboard
31 MiniMax-M3.1-Flash 1,504 1396 to 1628 Oct 2026 rotomarena.org Official leaderboard
32 gpt-oss-20b 1,500 Aug 2025 rotomarena.org Official leaderboard
33 Nemotron 3.5 Super VL 1,484 1380 to 1602 Oct 2026 rotomarena.org Official leaderboard
34 Perplexity Decider V1.2 closed weights 1,466 1355 to 1585 rotomarena.org Official leaderboard
35 Microsoft-Decision-1 closed weights 1,458 1353 to 1571 rotomarena.org Official leaderboard
36 Ornith 1.5 35B open weights 1,435 1334 to 1550 rotomarena.org Official leaderboard
37 Drex v2.0 open weights 1,379 1279 to 1499 rotomarena.org Official leaderboard
38 Clef open weights 1,370 1264 to 1492 rotomarena.org Official leaderboard
39 Ling 3.0 Tiny open weights 1,325 ±119 rotomarena.org Official leaderboard
40 Drex v1.5 open weights 1,297 1178 to 1422 rotomarena.org Official leaderboard
41 MiniCPM5-2B 1,291 ±119 Sep 2026 rotomarena.org Official leaderboard
42 Laya open weights 1,006 823 to 1173 rotomarena.org Official leaderboard
43 Solar Decide Flash closed weights 862 549 to 1055 rotomarena.org Official leaderboard
#ModelScore ±SettingDateLinkSource
1 Gemini 4 Argon 2,304 2038 to 2619 Sep 2026 rotomarena.org Official leaderboard
2 Gemini 3.8 Flash 2,047 1843 to 2228 Sep 2026 rotomarena.org Official leaderboard
3 GPT-6.1 Sol 2,003 1791 to 2178 Sep 2026 rotomarena.org Official leaderboard
4 Muse Wonder 1,954 1651 to 2203 Oct 2026 rotomarena.org Official leaderboard
5 MiMo-V2.6 Flash 1,819 1703 to 1944 Sep 2026 rotomarena.org Official leaderboard
6 DeepSeek V4.1 Flash 1,790 1663 to 1945 Sep 2026 rotomarena.org Official leaderboard
7 Gemma 4 31B IT 1,783 ±117.5 Apr 2026 rotomarena.org Official leaderboard
8 Muse Spark Open 1,778 1614 to 1930 Oct 2026 rotomarena.org Official leaderboard
9 Muse Spark 1.3 1,754 1641 to 1874 Sep 2026 rotomarena.org Official leaderboard
10 Claude Opus 5.5 1,752 1609 to 1905 Sep 2026 rotomarena.org Official leaderboard
11 DeepSeek-V4-Pro 1,731 1457 to 2086 Apr 2026 rotomarena.org Official leaderboard
12 Grok 4.7 1,731 1481 to 2088 Sep 2026 rotomarena.org Official leaderboard
13 GPT-6 Luna 1,729 1623 to 1849 Sep 2026 rotomarena.org Official leaderboard
14 Claude Haiku 5.5 1,728 1624 to 1846 Oct 2026 rotomarena.org Official leaderboard
15 Claude Sonnet 5.5 1,709 1557 to 1868 Sep 2026 rotomarena.org Official leaderboard
16 Solar Pro 4 1,664 1540 to 1798 Aug 2026 rotomarena.org Official leaderboard
17 Kimi K3 1,660 ±192 Jul 2026 rotomarena.org Official leaderboard
18 GLM-5.3-Flash 1,658 1556 to 1769 Aug 2026 rotomarena.org Official leaderboard
19 GLM-5.3 1,641 1498 to 1805 Aug 2026 rotomarena.org Official leaderboard
20 MiMo-V2.6 Pro 1,629 ±153.5 Sep 2026 rotomarena.org Official leaderboard
21 MiniMax-M3 1,626 1384 to 1889 Jun 2026 rotomarena.org Official leaderboard
22 Qwen3.8 Flash 1,581 1485 to 1697 Aug 2026 rotomarena.org Official leaderboard
23 Qwen 3.8 27B 1,553 1412 to 1700 Aug 2026 rotomarena.org Official leaderboard
24 Beam 1.0 1,531 1373 to 1740 Oct 2026 rotomarena.org Official leaderboard
25 Qwen 3.8 Max 1,528 1298 to 1812 prime Aug 2026 rotomarena.org Official leaderboard
26 Ling-3.1-flash 1,510 1225 to 1810 Oct 2026 rotomarena.org Official leaderboard
27 MiniMax-M3.1-Flash 1,504 1396 to 1628 Oct 2026 rotomarena.org Official leaderboard
28 gpt-oss-20b 1,500 Aug 2025 rotomarena.org Official leaderboard
29 Nemotron 3.5 Super VL 1,484 1380 to 1602 Oct 2026 rotomarena.org Official leaderboard
30 MiniCPM5-2B 1,291 ±119 Sep 2026 rotomarena.org Official leaderboard

Rankings by part

We read 5 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Wins, Losses and Win rate. MiMo-V2.6 Flash leads 1 of them, and 4 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Wins

30 models we track have a score here.

  1. 1 MiMo-V2.6 Flash ↑4 71.0
  2. 2 Muse Spark 1.3 ↑7 68.0
  3. 3 Claude Haiku 5.5 ↑11 67.0
  4. 4 Gemma 4 31B IT ↑3 66.0
  5. 5 GPT-6 Luna ↑8 66.0
  6. 6 GLM-5.3-Flash ↑12 62.0
  7. 7 gpt-oss-20b ↑21 54.0
  8. 8 Qwen3.8 Flash ↑14 49.0
  9. 9 DeepSeek V4.1 Flash ↓3 46.0
  10. 10 Solar Pro 4 ↑6 38.0

The best score, 2,304, costs $0.24 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Google DeepMind Z.ai Xiaomi Corp Cost frontier One model at each effort level
0 500 1,000 1,500 2,000 2,500 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Gemini 4 Argon Gemini 3.8 Flash MiMo-V2.6 Flash GLM-5.3-Flash 0 500 1,000 1,500 2,000 2,500 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Gemini 4 Argon Gemini 3.8 Flash

Source: Official Rotom Champ Bench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Argon 2,304 $0.24
Gemini 3.8 Flash 2,047 $0.03
GPT-6.1 Sol 2,003 $0.09
Muse Wonder 1,954 $0.26
MiMo-V2.6 Flash 1,819 $0.005
DeepSeek V4.1 Flash 1,790 $0.01
Gemma 4 31B IT 1,783 $0.005
Muse Spark Open 1,778 $0.07
Muse Spark 1.3 1,754 $0.007
Claude Opus 5.5 1,752 $0.27
DeepSeek-V4-Pro 1,731 $0.11
Grok 4.7 1,731 $0.30
GPT-6 Luna 1,729 $0.005
Claude Haiku 5.5 1,728 $0.008
Claude Sonnet 5.5 1,709 $0.08
Solar Pro 4 1,664 $0.009
Kimi K3 1,660 $0.07
GPT-6 Luna pro 1,660 $0.02
GLM-5.3-Flash 1,658 $0.003
GLM-5.3 1,641 $0.01
MiMo-V2.6 Pro 1,629 $0.02
MiniMax-M3 1,626 $0.11
Qwen3.8 Flash 1,581 $0.01
Qwen 3.8 27B 1,553 $0.03
Beam 1.0 1,531 $0.03
Qwen 3.8 Max prime 1,528 $0.25
Ling-3.1-flash 1,510 $0.03
MiniMax-M3.1-Flash 1,504
gpt-oss-20b 1,500
Nemotron 3.5 Super VL 1,484
MiniCPM5-2B 1,291

The best score, 2,304, costs $0.24 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Google DeepMind Z.ai Xiaomi Corp Cost frontier One model at each effort level
0 500 1,000 1,500 2,000 2,500 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Gemini 4 Argon Gemini 3.8 Flash MiMo-V2.6 Flash GLM-5.3-Flash 0 500 1,000 1,500 2,000 2,500 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Gemini 4 Argon Gemini 3.8 Flash

Source: Official Rotom Champ Bench leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Argon 2,304 $0.24
Gemini 3.8 Flash 2,047 $0.03
GPT-6.1 Sol 2,003 $0.09
Muse Wonder 1,954 $0.26
MiMo-V2.6 Flash 1,819 $0.005
DeepSeek V4.1 Flash 1,790 $0.01
Gemma 4 31B IT 1,783 $0.005
Muse Spark Open 1,778 $0.07
Muse Spark 1.3 1,754 $0.007
Claude Opus 5.5 1,752 $0.27
DeepSeek-V4-Pro 1,731 $0.11
Grok 4.7 1,731 $0.30
GPT-6 Luna 1,729 $0.005
Claude Haiku 5.5 1,728 $0.008
Claude Sonnet 5.5 1,709 $0.08
Solar Pro 4 1,664 $0.009
Kimi K3 1,660 $0.07
GPT-6 Luna pro 1,660 $0.02
GLM-5.3-Flash 1,658 $0.003
GLM-5.3 1,641 $0.01
MiMo-V2.6 Pro 1,629 $0.02
MiniMax-M3 1,626 $0.11
Qwen3.8 Flash 1,581 $0.01
Qwen 3.8 27B 1,553 $0.03
Beam 1.0 1,531 $0.03
Qwen 3.8 Max prime 1,528 $0.25
Ling-3.1-flash 1,510 $0.03
MiniMax-M3.1-Flash 1,504
gpt-oss-20b 1,500
Nemotron 3.5 Super VL 1,484
MiniCPM5-2B 1,291

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Own leaderboard 1 leader

Source: Official Rotom Champ Bench leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
1,250 1,500 1,750 2,000 2,250 2,500 2026 1,250 1,500 1,750 2,000 2,250 2,500 2026

Source: Official Rotom Champ Bench leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Games and long tasks

All 261 in Games and long tasks
  • ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6.2 Astra 100%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games. Gemini 2.5 Pro 3.90
  • MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks. Seed2.0 Pro 49%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress. Llama 4 Scout 7.00
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
  • Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (43 entries).
  • Pass Models scored 30 models we track, plus 13 other entries on its leaderboard.
  • Pass Newest score The newest is from October 2026, for MiniMax-M3.1-Flash.
  • Pass Frontier models 4 of the top 10 have a score, led by Claude Opus 5.5 (#5).
  • Note Headroom Scored as an Elo rating, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on rotomarena.org: 31 rows for models we track and 13 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check