Benchmark Almanac

Rotom Gen9 Bench

GPT-6 Luna 2,318 leads since October 2026
Not in model cards

Models draft six level 100 Pokémon and play the whole team in Generation 9 OverUsed singles. Match results determine their ratings.

The s4 table dated October 21st, 2026 rates 11 language models across 107 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 Showdown
Items from
Live play or votes
Refresh
Live
Run by
A host
In views
Games, Agents, Kept fresh
Strongest model tested
Claude Haiku 5.5 #20 on the capability index
Last model added
October 2026
Scores refreshed
November 20th, 2026
13
Entries on its own leaderboard, 8 models we track

GPT-6 Luna leads with 2,318 GPT-6 Luna leads with 2,318

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. 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 1,000.

  1. 1 GPT-6 Luna pro 2,318
  2. 2 Muse Spark 1.3 2,194
  3. 3 DeepSeek V4.1 Flash 2,095
  4. 4 Gemma 4 31B IT 2,082
  5. 5 GPT-6 Luna Decisions 2,072
  6. 6 Jev 1.13 1,996
  7. 7 MiMo-V2.6 Flash 1,993
  8. 8 Claude Haiku 5.5 1,856
  9. 9 Ornith 1.5 35B 1,661
  10. 10 Ling 3.0 Tiny 1,597
  11. 11 gpt-oss-20b 1,597
  12. 12 Bonsai 2 27B 1,500
  13. 13 MiniCPM5-2B 1,189

Source: Official Rotom Gen9 Bench leaderboard

#EntryScore ±SettingDateLinkSource
1 GPT-6 Luna 2,318 1911 to 2597 pro Sep 2026 rotomarena.org Official leaderboard
2 Muse Spark 1.3 2,194 1743 to 2499 Sep 2026 rotomarena.org Official leaderboard
3 DeepSeek V4.1 Flash 2,095 1685 to 2330 Sep 2026 rotomarena.org Official leaderboard
4 Gemma 4 31B IT 2,082 1755 to 2288 Apr 2026 rotomarena.org Official leaderboard
5 GPT-6 Luna Decisions closed weights 2,072 1774 to 2270 rotomarena.org Official leaderboard
6 Jev 1.13 closed weights 1,996 1702 to 2211 rotomarena.org Official leaderboard
7 MiMo-V2.6 Flash 1,993 1594 to 2249 Sep 2026 rotomarena.org Official leaderboard
8 Claude Haiku 5.5 1,856 1473 to 2117 Oct 2026 rotomarena.org Official leaderboard
9 Ornith 1.5 35B open weights 1,661 1416 to 1951 rotomarena.org Official leaderboard
10 Ling 3.0 Tiny open weights 1,597 1350 to 1871 rotomarena.org Official leaderboard
11 gpt-oss-20b 1,597 1361 to 1892 Aug 2025 rotomarena.org Official leaderboard
12 Bonsai 2 27B open weights 1,500 rotomarena.org Official leaderboard
13 MiniCPM5-2B 1,189 974 to 1535 Sep 2026 rotomarena.org Official leaderboard
#ModelScore ±SettingDateLinkSource
1 GPT-6 Luna 2,318 1911 to 2597 pro Sep 2026 rotomarena.org Official leaderboard
2 Muse Spark 1.3 2,194 1743 to 2499 Sep 2026 rotomarena.org Official leaderboard
3 DeepSeek V4.1 Flash 2,095 1685 to 2330 Sep 2026 rotomarena.org Official leaderboard
4 Gemma 4 31B IT 2,082 1755 to 2288 Apr 2026 rotomarena.org Official leaderboard
5 MiMo-V2.6 Flash 1,993 1594 to 2249 Sep 2026 rotomarena.org Official leaderboard
6 Claude Haiku 5.5 1,856 1473 to 2117 Oct 2026 rotomarena.org Official leaderboard
7 gpt-oss-20b 1,597 1361 to 1892 Aug 2025 rotomarena.org Official leaderboard
8 MiniCPM5-2B 1,189 974 to 1535 Sep 2026 rotomarena.org Official leaderboard

Number one over time

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

OpenAI
Own leaderboard 1 leader
GPT-6 Luna 2,318

Source: Official Rotom Gen9 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
1,000 1,500 2,000 2,500 2026 1,000 1,500 2,000 2,500 2026

Source: Official Rotom Gen9 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 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (13 entries).
  • Pass Models scored 8 models we track, plus 5 other entries on its leaderboard.
  • Pass Newest score The newest is from October 2026, for Claude Haiku 5.5.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Haiku 5.5 (#20).
  • Note Headroom Scored as an Elo rating, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on rotomarena.org: 8 rows for models we track and 5 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