gpt-5.5-pro
gpt-5.5-pro (OpenAI) scored 94.1% composite across 37 tasks - code, UI, full websites, SVG, marketing pages, dashboards, animations, and Australian legal and accounting. Graded by execution, and the visual builds by a cross-family vision panel (leave-one-family-out). Run on 2026-06-23.
Composite
94.1%
95% CI 89.9% - 97.4%
Est. cost
$55.6394
USD
Tokens
317,658
generation
Wall clock
6 min
end to end
Gate pass
-
validated
Separation
-
good vs bad
By domain
Composite score per domain, weakest first. Judge is the vision model’s read, shown for the visual domains.
svg-scene
1 task
58.3%
judge
58.3%
composite
Animation
2 tasks
64.6%
judge
64.6%
composite
Marketing page
1 task
77.1%
judge
77.1%
composite
Auth screen
1 task
83.3%
judge
83.3%
composite
Dashboard
1 task
85.4%
judge
85.4%
composite
Australian law
5 tasks
95.0%
composite
UI components
5 tasks
97.6%
composite
Australian accounting
6 tasks
98.3%
composite
Websites
3 tasks
98.7%
composite
Programming
6 tasks
100.0%
composite
SVG and graphics
6 tasks
100.0%
composite
Outputs
The actual rendered output. Open any tile to view it in a popup, or compare the same task across every model.
Open
Compare modelsANIM-0001judge 3.1/5
Open
Compare modelsANIM-0002judge 4.1/5
Open
Compare modelsAUTH-0001judge 4.3/5
Open
Compare modelsDASH-0001judge 4.4/5
Open
Compare modelsMKT-0001judge 4.1/5
Open
Compare modelsSCENE-0001judge 3.3/5
Open
Compare modelsSVG-0001100.0%
Open
Compare modelsSVG-0002100.0%
Open
Compare modelsSVG-0003100.0%
Open
Compare modelsSVG-0004100.0%
![gpt-5.5-pro SVG and graphics output - Tiny bar chart for values [3, 7, 5, 9]](/lab-thumbs/gpt-5-5-pro/SVG-0005.jpg)
Open
Compare modelsSVG-0005100.0%
Open
Compare modelsSVG-0006100.0%
Open
Compare modelsUI-000188.1%
Open
Compare modelsUI-0002100.0%
Open
Compare modelsUI-0003100.0%
Open
Compare modelsUI-000499.8%
Open
Compare modelsUI-0005100.0%
Open
Compare modelsWEB-0001100.0%
Open
Compare modelsWEB-0002100.0%
Open
Compare modelsWEB-000396.2%
Objective tasks
Programming, Australian legal and accounting, graded by execution. 15 of 17 scored a perfect 100.0%; the rest are below. Open the answer in a popup, or compare it across every model.