BREAKING
Grok 4.5 Debuts at No. 13
Big Jump From No. 29 to No. 13
Grok 4.3
29
Grok 4.5
13
0
%
Bash Recovery gain
0
%
Task success gain
0
Live sessions
0
$
Per M input tokens
0
$
Per M output tokens
0
tps
Generation speed
Strengths and Caveats
Strengths
●
64.7% SWE Bench Pro resolve
●
Strong on hallucination, 4th
●
Low supervision workflows
Caveats
●
Not top of every leaderboard
●
May trail Opus depth
●
Weekly token limits reported
A Cost-Focused Challenger
AI NEWS BLITZ
SpaceXAI's Grok 4.5 lands at number 13 on the Agent Arena leaderboard.