Models

Model set

Every model with an official Overall rank is shown by default. Partial results keep their rank; the icon explains what evidence is incomplete.

Reasoning view

Rank by

Reasoning view applies to Scores, Reasoning variants, Head-to-head, Per game heatmap, Economics (when available), and Cycle detector.

Compare two entries

Select two models to inspect their evidence together.

Rankings

Selected models, ordered by the Rank by choice above

Conservative score is the public leaderboard score after public-score adjustments; the other Rank by choices show diagnostic views when the selected models have that data.

Scores

Overall min-max and mean; small dots are each game (averaged across reasoning variants where scores exist)

Economics

Coming soon

Generation cost versus score will appear when the economics segment is available.

Overall

Reasoning variants

Per game

Strength heatmap (per-game scores averaged across reasoning variants)

Head-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.

More charts

Less-used views open on demand, so the page stays short.

Per-game comparison Top selected models within each game

Each panel ranks selected models within one game for the active reasoning view.

Cycle detector Non-transitive matchups between selected models

What 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.

Output tokens and API latency How much each model wrote, and how fast

These views will appear when the economics segment is available.

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
Claude Opus 5.5	Complete evidence	1	78.1	27.5	100.0
GPT-6 Astra	Complete evidence	2	77.9	38.0	100.0
Claude Fable 5.1	Complete evidence	3	74.5	35.8	95.8
GPT-6.1 Sol	Complete evidence	4	73.1	29.0	100.0
Claude Opus 5	Complete evidence	5	59.8	12.5	93.3
GPT-6 Sol	Complete evidence	6	57.6	8.5	90.0
Grok 4.7	Complete evidence	7	56.4	11.0	89.5
GPT-5.6 Sol	Complete evidence	8	49.4	9.6	78.5
GLM-5.3	Complete evidence	9	49.4	8.3	84.7
Claude Sonnet 5.5	Complete evidence	10	48.8	6.8	100.0
Claude Fable 5	Complete evidence	11	48.5	5.3	78.4
Muse Wonder	Complete evidence	12	46.8	1.2	82.4
Grok 4.6	Complete evidence	13	42.3	8.4	71.4
GPT-5.4	Complete evidence	14	41.9	0.6	100.0
Gemini 3.8 Flash	Complete evidence	15	41.3	4.4	76.5
Muse Spark 1.3	Complete evidence	16	40.9	10.5	74.1
Claude Haiku 5.5	Complete evidence	17	40.5	6.5	81.5
Grok 4.5	Complete evidence	18	40.0	0.8	72.2
GLM-5.3 Flash	Complete evidence	19	40.0	9.2	71.2
GLM-5.3 FlashX	Complete evidence	20	39.9	6.2	71.8
Claude Opus 4.8	Complete evidence	21	39.4	4.3	74.1
Hy4 Preview	Complete evidence	22	39.1	2.2	72.6
Qwen3.8 Max 0902	Complete evidence	23	38.1	5.6	82.1
Ox Alpha	Complete evidence	24	37.5	7.8	62.7
Qwen3.8 2.4T A95B	Complete evidence	25	37.3	7.3	74.0
Gemini 3.6 Flash	Complete evidence	26	36.6	2.5	74.3
DeepSeek V4.1 Flash	Complete evidence	27	36.2	6.1	73.0
GPT-5.5	Complete evidence	28	36.1	3.8	76.6
Ember-1	Complete evidence	29	35.6	7.6	64.2
Gemini 3.5 Flash	Complete evidence	30	35.2	4.2	77.1
Muse Spark 1.2	Complete evidence	31	34.9	5.7	68.7
Qwen3.8 Max	Complete evidence	32	33.8	5.4	79.0
Kimi K3	Complete evidence	33	33.6	0.9	87.1
Claude Opus 4.5	Complete evidence	34	33.4	7.5	57.0
GPT-6 Luna	Complete evidence	35	33.2	8.3	76.1
Gemini 3.1 Pro Preview	Complete evidence	36	32.1	7.1	70.5
Claude Sonnet 5	Complete evidence	37	32.0	6.3	59.0
DeepSeek V4 Pro	Complete evidence	38	31.7	4.2	69.5
GPT-5.6 Terra	Complete evidence	39	31.7	8.7	70.4
o3	Complete evidence	40	31.3	7.0	51.4
DeepSeek V4 Pro 0813	Complete evidence	41	31.0	0.5	67.1
Qwen3.8 27B	Coverage incomplete: generation quality is below the publication threshold	42	30.2	6.4	64.6
GPT-5	Coverage incomplete: one or more documented settings are untested or ineligible	43	29.6	0.0	63.9
GLM-5.2	Complete evidence	44	29.2	5.7	70.1
GPT-5.6 Luna	Complete evidence	45	27.4	0.0	72.4
Space Bunny Alpha	Complete evidence	46	26.3	6.2	62.7
Seed 2.0 Code	Complete evidence	47	25.8	4.2	68.3
Step 3.7 Flash	Complete evidence	48	24.3	7.5	47.5
DeepSeek V4 Flash	Complete evidence	49	24.1	4.8	72.1
Gemini 3.5 Flash Lite	Complete evidence	50	23.9	5.3	53.6
Nemotron 3 Ultra 550B A55B	Coverage incomplete: one or more documented settings are untested or ineligible	51	21.6	5.2	53.0
Muse Glimmer 30B	Complete evidence	52	21.5	7.0	44.0
GPT-OSS 120B	Complete evidence	53	21.4	6.7	49.5
Inkling Small	Complete evidence	54	21.1	6.1	50.7
Inkling	Complete evidence	55	20.7	6.6	49.4
Solar Mini 4	Coverage incomplete: the official generation cohort is incomplete; generation quality is below the publication threshold	56	17.5	5.3	42.7
Mercury 2.5 Preview	Complete evidence	57	16.8	6.5	38.6
Gemini 3.1 Flash Lite	Coverage incomplete: one or more documented settings are untested or ineligible	58	16.6	0.5	35.7

Charts

Chart controls