BREAKING
xAI launches Grok 4.5
0$/M
Input tokens
0$/M
Output tokens
0K
Context window
Terminal Bench 2.1 scores
Fable 584.3
GPT-5.583.4
Grok 4.583.3
Cost per task in agents
Grok 4.52.49
Fable 511.8
Planner vs implementer
Fable 5Orchestrator
Plans the work
Reviews changes
Higher cost
Grok 4.5Implementer
Executes edits
High-volume code
Low cost
Price-performance shift
AI NEWS BLITZ
xAI has released Grok 4.5, a coding-focused model that matches frontier rivals at a fraction of the cost.