GROK 4.5 UNDERCUTS RIVALS ON PRICE DESPITE BENCHMARK GAP

Grok 4.5 arrives with lower performance metrics than Fable 5 and GPT-5.5 on coding benchmarks, but xAI's pricing strategy may offset the gap. The model requires 4.2 times fewer tokens than Opus 4.8 to accomplish similar tasks, translating to substantial cost savings for users. At $2 per million input tokens, Grok 4.5 undercuts competitor pricing substantially. The model was trained using tens of thousands of Nvidia GB300 GPUs, positioning it as a competitive option for price-sensitive deployments. The release reflects an industry shift toward cost-efficiency. As AI model capabilities converge across major providers, pricing and efficiency metrics increasingly influence adoption decisions. EU availability is scheduled for mid-July, expanding access beyond existing markets.