Step 5 Preview 跑一次 Intelligence Index 每任务成本约 $0.72。同分模型成本高得多：Kimi K3（max）约 $2.00（2.8 倍），GLM-5.3（max）约 $2.01 且只多一分。Token 用量不是原因：Step 5 Preview 跑该 Index 约用 164M 输出 token，Kimi K3（max）约 170M。差异在定价，$1/$2.70 每 1M 输入/输出 token，对比 Kimi K3 的 $3/$15。MiMo-V2.6-Pro 是唯一得分更高（46）且每任务成本更低的模型（$0.13）。
Step 5 Preview costs ~$0.72 per task to run the Intelligence Index. Models at the same score cost considerably more: Kimi K3 (max) ~$2.00 (2.8x), and GLM-5.3 (max) ~$2.01 for one more point. Token usage is not the driver: Step 5 Preview uses ~164M output tokens to run the Index against ~170M for Kimi K3 (max). The difference is pricing, at $1/$2.70 per 1M input/output tokens against $3/$15 for Kimi K3. MiMo-V2.6-Pro is the only model scoring higher (46) at a lower cost per task ($0.13).
来源：@ArtificialAnlys · x.com