xAI has introduced Grok 4.5, a new language model developed using tens of thousands of Nvidia GB300 GPUs, according to The Decoder. The model is designed to reduce token consumption, using 4.2 times fewer tokens than Opus 4.8, which helps lower operational costs significantly.
While Grok 4.5 trails behind competitors like Fable 5 and GPT-5.5 in coding benchmarks, it offers a competitive edge by costing only $2 per million input tokens—significantly cheaper than many alternatives. The model’s availability in the European Union is expected around mid-July, The Decoder also reported.
For Japanese investors and traders, Grok 4.5’s improved cost efficiency and token usage could influence the adoption of AI-driven tools across FX, crypto, and equities markets, where quick, cost-effective data processing is increasingly valuable.