BREAKING
xAI Unveils Grok 4.5
Built for Coding and Agents
1
Data curation
↓
2
Large-scale RL
↓
3
Cursor tuning
↓
4
Grok Build default
0
Blackwell GPUs
0
Grace CPUs
0
x
vs Hopper output
Benchmark Results
Terminal-Bench 2.1
83.3
SWE-Bench Pro
64.7
DeepSWE 1.0
62
0
TPS
Inference speed
0
$
Input /M tokens
0
$
Output /M tokens
First Model Citing GB300
AI NEWS BLITZ
xAI has revealed Grok 4.5, a new model built for coding and agentic tasks.