Grok 4.5 DeepSWE 得分 53% —— 比 Fable 5 便宜 90%
xAI 發布了 Grok 4.5，價格為每百萬 tokens 2/6 美元，顯著低於 Fable 5（10/50 美元）等競爭對手。該定價策略瞄準了一個市場轉變，即效率比基準測試的微小差距更重要。
xAI released Grok 4.5, a model targeting coding and agentic workflows, with a pricing structure of $2 input and $6 output per million tokens, drastically undercutting competitors like Fable 5 ($10/$50) and Opus 4.8 ($5/$25). benchmark results are competitive but variable: Grok 4.5 scores 83.3% on Terminal Bench 2.1, nearly matching GPT 5.5, but falls behind on DeepSWE 1.1 with 53% compared to Fable 5's 70%. xai claims the model uses 4.2 times fewer tokens than Opus 4.8 on SWE Bench Pro, positioning it as the most efficient option in its performance tier despite raw accuracy gaps. available now via Grok Build and Cursor, excluding the EU, this release signals a price war where vendors trade minor accuracy losses for massive cost savings.