Grok 4.5：トークン効率2倍でOpus 4.7のコストを76%削減
spacexaiが「opusクラス」モデル「grok 4.5」をリリース。100万トークンあたり$2/$6の価格設定でanthropicやopenaiを下回る。この攻撃的な価格設定は、効率性とコスト覇権を巡る新たな競争の幕開けを示している。
spacexai released grok 4.5, marketing it as a high-performance model capable of coding, research, and office work, with ceo elon musk claiming it rivals anthropic opus 4.7 in capability. the model distinguishes itself through economics, costing $2 for input and $6 for output tokens per million, which is approximately 76% cheaper than opus 4.7's output costs. this move pressures competitors like openai and anthropic, who are holding their ground on higher pricing tiers despite the looming release of gpt 5.6. by prioritizing token efficiency and cost reduction, spacexai aims to capture enterprise market share and redefine the value proposition for deploying high-end ai.