BREAKING
xAI Tests 1.5T Grok 4.5 Internally
0
T
parameters
0
%
size increase
0
T
Grok 4.4 params
Grok 4.5 vs Claude Opus
Grok 4.5
V9 base
●
1.5T parameters
●
Cursor coding data
Claude Opus
Anthropic
●
Performance benchmark
●
Claimed equal or below
Deployed First at SpaceX and Tesla
1
V9 foundation
↓
2
Cursor data
↓
3
SpaceX + Tesla
↓
4
Real-world RL
New Models Every Month in 2026
Benchmarks Still Undisclosed
AI NEWS BLITZ
xAI has launched its new 1.5-trillion-parameter Grok 4.5 in a private beta.