Compare two entries
Select two models to inspect their evidence together.
Reasoning view
Applies to Scores, Reasoning variants, Head-to-head, Per game heatmap, Economics (when available), and Cycle detector.
Rank by
Diagnostic orderingConservative score is the public leaderboard score after public-score adjustments; other buttons show diagnostic views when the selected models have that data.
Economics
Coming soonGeneration cost versus score will appear when the economics segment is available.
Generation cost evidence, output-token counts, API latency, and scores include only economics-eligible games once that segment is available. Cost means equal-weight unique model/game/reasoning cells; hover text gives the median, observed-run count, and evidence basis. Output tokens and API latency are generated-code/repair response telemetry; latency excludes local validation and compile probe time. “All settings” combines public reasoning variants; the three bracket controls select their model-relative provider settings.
Scores
Overall min-max and mean; small dots are each game (averaged across reasoning variants where scores exist)Overall
Reasoning variantsHead-to-head
Combined view across tested provider settings.Off-diagonal cells use win rate for the row model against the column opponent (W-L-D in the tooltip). Use Reasoning view to show one reasoning variant.
Per game
Strength heatmap (per-game scores averaged across reasoning variants)Per-game comparison
Top selected models within each gameShow per-game comparison panels
Each panel ranks selected models within one game for the active reasoning view.
Cycle detector
Non-transitive matchupsWhat it finds: Leaderboard averages can hide matchup loops. The detector searches the currently selected models and reasoning view for cases where A outperforms B, B outperforms C, and C outperforms A.
How to read it: In A → B → C → A, each arrow points from the model with the higher pair score to the model it outscored. Hover over a result for each edge’s win-loss-draw record and pair score.
Each arrow requires at least four matches and a pair score above 50%, with a draw counting as half a win. Results are ordered by the cycle’s weakest edge: weakest edge +x pp is the smallest of the three margins above 50%, while n is the total match count across all three edges. The detector shows up to ten cycles. These are descriptive matchup patterns, not tests of statistical significance. “No cycles” means none met these rules in the current selection and reasoning view; it does not prove that the matchups are transitive.
Static rankings — current page release
This table remains available without chart scripts. It describes this page's release; restoring a requested comparison and its filters requires JavaScript. Open benchmark rankings for the detailed tables.
Current comparison tables
Overall leaderboard scores
All officially ranked Overall models from the same data as the interactive charts. Partial rows keep their official rank and show that their evidence is incomplete.
| Model | Evidence status | Rank | Mean | Min | Max |
|---|---|---|---|---|---|
 GPT-6 Astra | Complete evidence | 1 | 84.0 | 50.8 | 100.0 |
 Claude Fable 5.1 | Complete evidence | 2 | 80.7 | 33.6 | 100.0 |
| Claude Opus 5.5 | Complete evidence | 3 | 72.3 | 33.3 | 92.6 |
| Claude Opus 5 | Complete evidence | 4 | 63.6 | 15.4 | 90.6 |
 GLM 5.3 | Coverage incomplete: the official generation cohort is incomplete | 5 | 59.3 | 13.8 | 97.7 |
| GPT-5.6 Sol | Complete evidence | 6 | 56.4 | 9.9 | 89.0 |
| GPT-6 Sol | Complete evidence | 7 | 56.1 | 10.7 | 83.3 |
 Grok 4.7 | Coverage incomplete: the official generation cohort is incomplete | 8 | 55.2 | 11.0 | 79.8 |
| Claude Fable 5 | Complete evidence | 9 | 54.6 | 4.5 | 88.4 |
| Grok 4.6 | Complete evidence | 10 | 48.1 | 12.9 | 81.7 |
 MiMo-V2.6-Pro | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing; the official generation cohort is incomplete | 11 | 47.4 | 4.5 | 77.5 |
| GPT-5.4 | Complete evidence | 12 | 46.5 | 2.1 | 100.0 |
| GLM 5.3 Flash | Complete evidence | 13 | 46.4 | 9.4 | 74.2 |
 Gemini 3.8 Flash | Complete evidence | 14 | 46.0 | 8.6 | 83.6 |
| Grok 4.5 | Complete evidence | 15 | 45.4 | 0.0 | 79.3 |
 Hy4 Preview | Complete evidence | 16 | 45.3 | 10.5 | 81.4 |
 Muse Spark 1.3 | Coverage incomplete: the official generation cohort is incomplete | 17 | 45.3 | 12.8 | 77.7 |
| Claude Opus 4.8 | Complete evidence | 18 | 43.4 | 9.4 | 77.3 |
 Qwen3.8 2.4t A95B | Complete evidence | 19 | 42.8 | 12.6 | 82.7 |
| Gemini 3.6 Flash | Complete evidence | 20 | 42.4 | 6.5 | 84.8 |
| Ox Alpha | Complete evidence | 21 | 42.1 | 13.3 | 66.9 |
| GLM-5.3 FlashX | Complete evidence | 22 | 41.7 | 6.3 | 69.7 |
| Gemini 3.5 Flash | Coverage incomplete: fixed opponent-and-seat coverage is incomplete | 23 | 40.5 | 3.2 | 83.1 |
| MiMo-V2.6-Flash | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 24 | 40.5 | 10.6 | 74.8 |
| Qwen3.8 Max 0902 | Coverage incomplete: the official generation cohort is incomplete | 25 | 40.5 | 0.4 | 78.0 |
| GPT-5.5 | Complete evidence | 26 | 40.3 | 10.6 | 83.9 |
 Kimi K3 | Coverage incomplete: the official generation cohort is incomplete | 27 | 40.1 | 10.7 | 88.2 |
| GPT 4.1 Nano | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing; the official generation cohort is incomplete | 28 | 40.1 | 40.1 | 40.1 |
| Muse Spark 1.2 | Coverage incomplete: the official generation cohort is incomplete | 29 | 39.7 | 10.3 | 74.0 |
 DeepSeek V4.1 Flash | Complete evidence | 30 | 39.5 | 8.8 | 77.6 |
| Qwen3.8 Max | Complete evidence | 31 | 38.0 | 11.6 | 90.9 |
| Claude Sonnet 5 | Complete evidence | 32 | 37.6 | 13.3 | 64.3 |
| Claude Opus 4.5 | Complete evidence | 33 | 37.0 | 9.4 | 66.0 |
| Qwen3.8 Flash | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 34 | 36.7 | 8.3 | 73.0 |
| GPT-5.6 Terra | Complete evidence | 35 | 35.3 | 8.3 | 75.5 |
| Qwen3.8 27B | Coverage incomplete: generation quality is below the publication threshold; fixed opponent-and-seat coverage is incomplete; the official generation cohort is incomplete | 36 | 35.0 | 18.2 | 74.8 |
| DeepSeek V4 Pro | Complete evidence | 37 | 34.9 | 7.9 | 72.9 |
| DeepSeek V4 Pro 0813 | Complete evidence | 38 | 34.9 | 0.0 | 72.3 |
| GLM-5.2 | Complete evidence | 39 | 33.7 | 8.1 | 74.8 |
| MiMo-V2.5-Pro | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 40 | 33.3 | 3.0 | 67.0 |
| GPT 5 | Coverage incomplete: one or more documented settings are untested or ineligible; fixed opponent-and-seat coverage is incomplete | 41 | 33.1 | 0.7 | 70.2 |
| GPT-6 Luna | Coverage incomplete: the official generation cohort is incomplete | 42 | 32.4 | 11.2 | 78.8 |
| Qwen3.7 Max | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 43 | 32.3 | 9.9 | 72.6 |
| GPT 4.1 | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 44 | 30.7 | 6.4 | 56.5 |
| GPT-5.6 Luna | Complete evidence | 45 | 30.5 | 4.2 | 78.9 |
| O3 | Coverage incomplete: generation quality is below the publication threshold | 46 | 30.4 | 7.4 | 53.0 |
| Kimi K2.7 Code | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 47 | 30.2 | 10.1 | 52.1 |
| GPT-5.4 Mini | Coverage incomplete: one or more bracket scores are missing; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 48 | 29.3 | 0.0 | 59.8 |
 Seed 2.0 Code | Complete evidence | 49 | 29.3 | 7.8 | 74.7 |
| GPT-5.4 Nano | Coverage incomplete: one or more bracket scores are missing; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 50 | 27.9 | 9.8 | 57.8 |
| DeepSeek V4 Flash | Complete evidence | 51 | 27.5 | 8.1 | 79.2 |
| Laguna S 2.1 | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 52 | 27.3 | 6.9 | 48.8 |
| Gemini 3.5 Flash Lite | Complete evidence | 53 | 26.8 | 9.3 | 54.8 |
| Hy3 Preview | Coverage incomplete: one or more bracket scores are missing; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 54 | 26.5 | 3.9 | 61.5 |
| Grok Build 0.1 | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 55 | 26.3 | 12.3 | 52.6 |
| Qwen3.7 Plus | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 56 | 25.9 | 9.0 | 71.6 |
| LongCat 2.0 | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 57 | 25.1 | 7.2 | 42.2 |
| GPT-OSS 120B | Complete evidence | 58 | 24.7 | 6.3 | 52.3 |
 Step 3.7 Flash | Coverage incomplete: generation quality is below the publication threshold | 59 | 24.6 | 9.8 | 48.8 |
| Muse Glimmer 30B | Complete evidence | 60 | 24.1 | 6.2 | 48.9 |
| Gemma 4 31B | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing; the official generation cohort is incomplete | 61 | 23.9 | 13.6 | 50.7 |
 Nemotron 3 Ultra 550B A55B | Coverage incomplete: one or more documented settings are untested or ineligible | 62 | 23.3 | 4.9 | 54.9 |
| Inkling Small | Complete evidence | 63 | 23.3 | 3.6 | 57.1 |
 Minimax M3 | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 64 | 23.1 | 0.0 | 41.7 |
| Inkling | Coverage incomplete: generation quality is below the publication threshold | 65 | 22.8 | 8.5 | 61.8 |
| MiMo-V2.5 | Coverage incomplete: three distinct provider settings are not documented; one or more documented settings are untested or ineligible | 66 | 22.3 | 6.9 | 49.2 |
 Mistral Medium 3.5 | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing | 67 | 21.3 | 3.9 | 54.3 |
| North Mini Code | Coverage incomplete: one or more bracket scores are missing; the provider capability record is unavailable; generation-quality evidence is missing; generation quality is below the publication threshold; fixed opponent-and-seat coverage is incomplete | 68 | 20.6 | 13.7 | 49.7 |
| Mercury 2.5 Preview | Complete evidence | 69 | 19.7 | 9.8 | 39.9 |
 Nex N2 Pro | Coverage incomplete: one or more bracket scores are missing; three distinct provider settings are not documented; one or more documented settings are untested or ineligible; generation-quality evidence is missing; generation quality is below the publication threshold | 70 | 19.1 | 0.0 | 64.4 |
 Ling 3.0 Flash | Coverage incomplete: one or more bracket scores are missing; the provider capability record is unavailable; generation-quality evidence is missing; generation quality is below the publication threshold; fixed opponent-and-seat coverage is incomplete | 71 | 16.6 | 4.0 | 38.8 |
| Gemini 3.1 Flash Lite | Coverage incomplete: one or more documented settings are untested or ineligible | 72 | 16.4 | 0.8 | 31.7 |
| Granite 4.2 8B | Coverage incomplete: one or more bracket scores are missing; one or more documented settings are untested or ineligible; generation-quality evidence is missing; generation quality is below the publication threshold | 73 | 10.5 | 1.1 | 22.5 |