Grok 4.7 输出贵至 ¥40/M token,被便宜 16 倍的 MiMo V2.6 贴身对比
歸藏的AI工具箱 · wechat · 2026-09-22
Two same-day releases—Grok 4.7 and Xiaomi's MiMo V2.6—score similarly on third-party benchmarks but differ wildly on price: Grok 4.7 costs ¥3.35 input / ¥40 output per million tokens, while MiMo V2.6 Pro is ¥0.025 (cache hit) / ¥6, and Flash output is ¥2, undercutting even DeepSeek V4's trough pricing. Early Grok 4.7 feedback complains it's expensive and slow. The author argues the model market now only rewards the cheapest or the frontier, and suspects both releases rushed ahead of a possible OpenAI GPT-6 launch.
所属事件:Grok 4.7 与小米 MiMo-V2.6 同夜对狙(5 条相关)→
「模型」频道最新
- 11 个 LLM 剑桥辩论社对决 1168 场：Claude Fable 5.2 压倒性夺冠 — AymericRoucher · 2026-11-12
- 用户实测 Claude App：widget 模块看不到用户编辑、移动端也不显示 — petergyang · 2026-11-12
- Allen-Zhu新论文：Canon层随机卷积权重也能提升模型推理 — teortaxesTex · 2026-11-12
- Ricursive 发布开源权重 Ripple-65B：TerminalBench 称霸美国开源榜 — StanfordAILab · 2026-11-12
- 被严重低估的资源：跨模型跨厂商的工具调用可靠性实测 — ricklamers · 2026-11-12
- Datalab 快速模式 OCR 准确率超 Opus 5.5，价格仅 1/4 到 1/68 — VikParuchuri · 2026-11-12