Thinking Machines 发布 Inkling 2 Small：594B MoE 模型，68 小时完成预训练
原标题：[AINews] Inkling 2 Small: 17A-594B trained in 68 hours on 12T tokens, beating Muse Spark
AI 摘要
Thinking Machines 发布 Inkling 2 Small，这是一个总参数 594B、激活参数 17B 的多模态 MoE 模型，在 GB300 上以原生 NVFP4 精度用 68 小时完成预训练，主预训练阶段消耗 8.4T tokens。模型已在 Tinker 平台上线，Apache 2.0 权重承诺在下周外部安全测试完成后开放。独立评测显示其得分较前代大幅提升，被视为当前中国以外得分最高的开放权重模型之一，但输出 token 消耗约为前代的 5 倍。
正文节选
There was a time in the early days of Thinking Machines, post 2025 start and cofounder departures, when their going concern as a neolab was very much in doubt. Those days are clearly behind us, as Tinker (now a year old!) and Interaction Models and Inkling and other high-impact contributions to AI research have converted Thinky’s highly credentialed potential into actual concrete ships that establish that they are now a high-functioning neolab at the US frontier. Anyway, here’s Inkling 2 Small -