grok-4.20-reasoning
grok-4.20-reasoning scored 71.4% composite across 87 tasks. This run covers Australian accounting, Animation, Auth screens, Dashboards, Australian law, Marketing pages, Programming, SVG and graphics, svg-scene, UI components, Websites. Published 2026-06-25.
API single-shotdefault reasoning
Composite
71.4%
95% CI 62.6% - 79.3%
Est. cost
$0.4885
USD
Tokens
219,591
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
41.7%
judge
41.7%
composite
Websites
15 tasks
45.0%
composite
Animation
2 tasks
58.3%
judge
58.3%
composite
Auth screen
1 task
66.7%
judge
66.7%
composite
Australian accounting
23 tasks
67.2%
composite
Marketing page
1 task
68.8%
judge
68.8%
composite
Dashboard
1 task
70.8%
judge
70.8%
composite
UI components
19 tasks
73.6%
composite
Australian law
5 tasks
80.0%
composite
SVG and graphics
13 tasks
94.2%
composite
Programming
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.
53 results
Objective tasks
Programming, Australian legal and accounting, graded by execution. 21 of 34 scored a perfect 100.0%; the rest are below. Open the answer in a popup, or compare it across every model.