BREAKING
Grok 4.5 Hits No. 3 on Code Arena
A Leap From No. 62 to No. 3
Grok 4.3
62
Grok 4.5
3
0
%
SWE Bench Pro
0
%
Terminal Bench 2.1
0
TPS
Speed
0
$
Input /1M
0
$
Output /1M
0
tok
Context
Strong Front-End, Weaker Polish
Strengths
●
Low hallucination rate
●
Strong planning, less slop
●
Passes Playwright tests
Gaps
●
Trails Opus and Fable on UI
●
Leans front-end over back-end
Available on Cursor and API
AI NEWS BLITZ
xAI's new coding model Grok 4.5 has climbed to number three on Code Arena Frontend.