非文本AI模型Jev发布数周后，其开发商TypeSafe估值达7.5亿美元。该模型宣称比LLM更快、用更少token。
TypeSafe secured a $7.5B valuation for Jev within weeks of its launch—based not on revenue or users, but on demonstrable efficiency gains: markedly faster inference and drastically lower token consumption versus LLMs. This signals investor appetite for architectural alternatives that directly reduce operational cost per task. For early-stage AI teams, it underscores a strategic pivot: value is increasingly tied to resource efficiency—not scale alone. When customers pay per agent invocation or deploy on constrained edge hardware, cutting token usage by 40–60% can slash cloud spend, accelerate time-to-POC, and widen the margin for enterprise contracts. Founders should prioritize measurable cost-per-action metrics in technical design and pitch decks over abstract capability claims.