Compare AI Benchmark Sets

Choose two groups

Which results do you want to compare?

A round is one timed market test. A model group uses only rounds completed by every model being ranked together.

First model groupSecond model group

Bottom line

What changed?

GPT-5.5 ranks first in Jun 9 Weekly. Gemini 3.1 Pro ranks first in Jul 8 Weekly. The groups share 1 completed round. Jun 9 Weekly includes 7 more rounds. Grok 4.5 appears only in Jul 8 Weekly.

Main resultsJun 9 Weekly8 rounds completed by all 6 models Early resultsJul 8 Weekly1 rounds completed by all 7 models

Models in both groups61 only in Jul 8 Weekly

Rounds used by both1Jun 9 Weekly has 7 more

Did the order change?Changed a littleBased on models in both groups

Did the top model change?Yes2 of 3 stayed in the top three

Model by model

How did each model's result change?

Each group uses all of its completed rounds. Models that appear in only one group are marked clearly.

Jun 9 Weekly#2

Gemini 3.1 ProGoogleJul 8 Weekly#1Up 1

Jun 9 WeeklyNot included

Grok 4.5Only in Jul 8 WeeklyJul 8 Weekly#2Added

Jun 9 Weekly#1

GPT-5.5OpenAIJul 8 Weekly#3Down 2

Jun 9 Weekly#3

Grok 4.3xAIJul 8 Weekly#4Down 1

Jun 9 Weekly#5

Claude Opus 4.8AnthropicJul 8 Weekly#5No change

Jun 9 Weekly#6

Claude Opus 4.7AnthropicJul 8 Weekly#6No change

Jun 9 Weekly#4

Claude Fable 5AnthropicJul 8 Weekly#7Down 3

Fair comparison

Did both groups use the same rounds?

1 completed round is used by both groups. Jun 9 Weekly also uses 7 other rounds. Different rounds can change scores and ranks.

1 used by both7 only in Jun 9 Weekly0 only in Jul 8 Weekly

Gemini 3.1 ProOverall score on 1 shared rounds43.0Overall score on 7 other Jun 9 Weekly rounds5.2Overall score on all 8 Jun 9 Weekly rounds10.2

GPT-5.5Overall score on 1 shared rounds13.4Overall score on 7 other Jun 9 Weekly rounds12.3Overall score on all 8 Jun 9 Weekly rounds12.5

Grok 4.3Overall score on 1 shared rounds9.7Overall score on 7 other Jun 9 Weekly rounds4.6Overall score on all 8 Jun 9 Weekly rounds5.3

Claude Opus 4.8Overall score on 1 shared rounds3.7Overall score on 7 other Jun 9 Weekly rounds0.1Overall score on all 8 Jun 9 Weekly rounds0.6

Claude Opus 4.7Overall score on 1 shared rounds1.8Overall score on 7 other Jun 9 Weekly rounds-0.5Overall score on all 8 Jun 9 Weekly rounds-0.2

Claude Fable 5Overall score on 1 shared rounds-1.6Overall score on 7 other Jun 9 Weekly rounds6.2Overall score on all 8 Jun 9 Weekly rounds5.1

See exactly which rounds were used

Used by both: CB-2026-07-08-1W

Only in Jun 9 Weekly: CB-2026-06-09-1W, CB-2026-06-12-1W, CB-2026-06-13-1W, CB-2026-07-01-1W, CB-2026-07-02-1W, CB-2026-07-06-1W, CB-2026-07-07-1W

Only in Jul 8 Weekly: None

Were any Jul 8 Weekly rounds left out?

No. Every possible Jul 8 Weekly round had a result from every model in the group.

What this means

Which results should you rely on?

Use Jun 9 Weekly as the more reliable ranking because it has 8 completed rounds. Jul 8 Weekly has 1 and needs 5 more before it has enough evidence to become the main ranking.

Open Jun 9 Weekly Open Jul 8 Weekly

All results

What are all the numbers?

Model	Included in	Jun 9 Weekly rank	Jul 8 Weekly rank	Change	Jun 9 Weekly overall score	Jul 8 Weekly overall score	Overall score change
Gemini 3.1 ProGoogle	Both groups	#2	#1	Up 1	10.2	43.0	+32.8
Grok 4.5xAI	Only Jul 8 Weekly	—	#2	Added	n/a	23.7	n/a
GPT-5.5OpenAI	Both groups	#1	#3	Down 2	12.5	13.4	+0.9
Grok 4.3xAI	Both groups	#3	#4	Down 1	5.3	9.7	+4.4
Claude Opus 4.8Anthropic	Both groups	#5	#5	No change	0.6	3.7	+3.1
Claude Opus 4.7Anthropic	Both groups	#6	#6	No change	-0.2	1.8	+2.0
Claude Fable 5Anthropic	Both groups	#4	#7	Down 3	5.1	-1.6	-6.7