BREAKING
SpaceXAI Launches Grok 4.5
0T
parameters (V9)
0$/1M
input tokens
0$/1M
output tokens
0tok/s
generation speed
Efficiency Over Raw Scores
Grok 4.5vendor claim
Opus-class or better
Edge in speed and cost
4.2x fewer tokens
Opus 4.8 Max
Reference model
Higher token use
Comparison baseline
From Private Beta to Public
1Beta from Jun 28
2Public Jul 8
3Grok Build, Cursor
4xAI API, console
Benchmarks Still Unverified
Monthly Models Through 2026
AI NEWS BLITZ
SpaceXAI has rolled out Grok 4.5, a new model built for coding and agentic work.