Champions BSS · own teams (c4) · 852 rated games · updated 2026-10-17
Champions BSS · own teams (c4)
Archived. A superseded version, kept for the record; its games count on no current leaderboard. The current one is Rotom Champ Bench.
The fourth Champions version (retired 13 Oct): models saw the last few turns instead of the whole battle log, kept no notes from turn to turn, built one team instead of three, and thought within 2,048 tokens.
Ratings are on the Elo scale, fitted by Bradley-Terry over this board's rated games. Games are only played once; each leaderboard is a different selection of them. GPT-OSS 20B is fixed at 1500 by definition; every other number is relative to it.3Models with fewer than 16 games are marked provisional. Click a column to sort.
| # | Model | Rating (95% CI)1 | Games | W | L | Win rate | Halluc.2 | Tokens / turn | Cost / game |
|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT-6 Luna Pro | 75 | 64 | 11 | 85% | 0.0% | 1,380 | $0.016 | |
| 2 | Gemini 3.8 Flash | 75 | 63 | 12 | 84% | 5.8% | 286 | $0.030 | |
| 3 | Claude Opus 5.5 | 15 | 9 | 6 | 60% | 0.0% | 714 | $0.266 | |
| 4 | GPT-6 Luna | 75 | 56 | 19 | 75% | 0.0% | 369 | $0.005 | |
| 5 | GPT-6 Luna Decisions | 69 | 51 | 18 | 74% | 0.0% | 0 | $0.003 | |
| — | PokéChamp | 66 | 47 | 19 | 71% | — | — | — | |
| 6 | DeepSeek V4.1 Flash | 69 | 48 | 21 | 70% | 2.2% | 1,109 | $0.012 | |
| 7 | Muse Spark 1.3 | 69 | 47 | 22 | 68% | 0.1% | 1,590 | $0.008 | |
| 8 | Mercury Decide | 69 | 46 | 23 | 67% | 0.0% | 0 | $0.001 | |
| 9 | Gemma 4 31B | 69 | 39 | 30 | 56% | 3.3% | 754 | $0.005 | |
| 10 | Claude Sonnet 5.5 | 15 | 4 | 11 | 27% | 0.0% | 447 | $0.082 | |
| 11 | Claude Haiku 5.5 | 69 | 37 | 32 | 54% | 1.3% | 926 | $0.009 | |
| 12 | Microsoft-Decision-1 | 69 | 37 | 32 | 54% | 0.0% | 0 | $0.001 | |
| 13 | GLM 5.3 Flash | 69 | 34 | 35 | 49% | 0.5% | 201 | $0.004 | |
| 14 | Perplexity Decider V1.1 27B | 69 | 34 | 35 | 49% | 0.0% | 0 | $0.001 | |
| 15 | GPT-OSS 20B | 75 | 34 | 41 | 45% | 3.6% | 1,279 | free | |
| 16 | Ornith 1.5 35B | 69 | 32 | 37 | 46% | 4.0% | 2,164 | free | |
| 17 | Solar Pro 4 | 69 | 30 | 39 | 44% | 0.7% | 1,781 | $0.009 | |
| 18 | Drex v1.5 | 69 | 27 | 42 | 39% | 10.0% | 0 | free | |
| 19 | Ling 3.0 Tiny | 69 | 25 | 44 | 36% | 2.8% | 2,352 | free | |
| 20 | Bonsai 2 27B | 69 | 24 | 45 | 35% | 1.3% | 1,784 | free | |
| 21 | Jev 1.13 | 69 | 24 | 45 | 35% | 0.0% | 0 | $0.002 | |
| 22 | MiMo V2.6 Flash | 69 | 21 | 48 | 30% | 2.4% | 855 | $0.006 | |
| — | Chimps at the Zoo | 66 | 10 | 56 | 15% | — | — | — | |
| 23 | MiniCPM5 2B | 69 | 6 | 63 | 9% | 3.4% | 1,607 | free | |
| 24 | Laya | 69 | 3 | 66 | 4% | 2.0% | 0 | free |
Rules
Every game records the rules it was played under; a board only rates games that match their version's stored rules.
| Version | Format | Teams | Bring | Thinking per turn | Tokens per turn | Prompt |
|---|---|---|---|---|---|---|
| c4 | gen9championsrotomarenabssregmb | own (drafted) | 3 of 6 | 2,048 tokens | 4,096 | c6b6787ac85f · draft b0a500bd511d |
Head to head
Wins–losses of the row model against the column model. Click a record for its games and replays.
| GPT-6 Luna Pro | Gemini 3.8 Flash | Claude Opus 5.5 | GPT-6 Luna | GPT-6 Luna Decisions | PokéChamp | DeepSeek V4.1 Flash | Muse Spark 1.3 | Mercury Decide | Gemma 4 31B | Claude Sonnet 5.5 | Claude Haiku 5.5 | Microsoft-Decision-1 | GLM 5.3 Flash | Perplexity Decider V1.1 27B | GPT-OSS 20B | Ornith 1.5 35B | Solar Pro 4 | Drex v1.5 | Ling 3.0 Tiny | Bonsai 2 27B | Jev 1.13 | MiMo V2.6 Flash | Chimps at the Zoo | MiniCPM5 2B | Laya | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GPT-6 Luna Pro | ||||||||||||||||||||||||||
| Gemini 3.8 Flash | ||||||||||||||||||||||||||
| Claude Opus 5.5 | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | ||||||
| GPT-6 Luna | ||||||||||||||||||||||||||
| GPT-6 Luna Decisions | · | · | ||||||||||||||||||||||||
| PokéChamp | · | · | · | |||||||||||||||||||||||
| DeepSeek V4.1 Flash | · | · | ||||||||||||||||||||||||
| Muse Spark 1.3 | · | · | ||||||||||||||||||||||||
| Mercury Decide | · | · | ||||||||||||||||||||||||
| Gemma 4 31B | · | · | ||||||||||||||||||||||||
| Claude Sonnet 5.5 | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | · | ||||||
| Claude Haiku 5.5 | · | · | ||||||||||||||||||||||||
| Microsoft-Decision-1 | · | · | ||||||||||||||||||||||||
| GLM 5.3 Flash | · | · | ||||||||||||||||||||||||
| Perplexity Decider V1.1 27B | · | · | ||||||||||||||||||||||||
| GPT-OSS 20B | ||||||||||||||||||||||||||
| Ornith 1.5 35B | · | · | ||||||||||||||||||||||||
| Solar Pro 4 | · | · | ||||||||||||||||||||||||
| Drex v1.5 | · | · | ||||||||||||||||||||||||
| Ling 3.0 Tiny | · | · | ||||||||||||||||||||||||
| Bonsai 2 27B | · | · | ||||||||||||||||||||||||
| Jev 1.13 | · | · | ||||||||||||||||||||||||
| MiMo V2.6 Flash | · | · | ||||||||||||||||||||||||
| Chimps at the Zoo | · | · | · | |||||||||||||||||||||||
| MiniCPM5 2B | · | · | ||||||||||||||||||||||||
| Laya | · | · |
- 95% intervals come from 1,000 parametric bootstrap refits. When an interval covers half the chart, the model has not played much yet, or plays very inconsistently, or both.
- Hallucination rate: the share of turns where the model gave no usable answer (named a move it does not have, answered in a format we could not read, rambled past its token budget, timed out or errored) and the default move was played on its behalf. Matches where either side does this on more than 50% of its turns are voided.
- Ratings need a reference point. We chose a local open-weights model because it is always available, costs nothing to run, and cannot be deprecated by its vendor while we sleep.