Original Summary

Where Jev and other Jev-style models like OpenAI&#x27;s Decisions and Cloudflare&#x27;s Clef are multimodal, supporting text or images in, decision out, SPEV natively supports audio and video inputs. These native capabilities allow SPEV to reason directly from audio without the lossy speech to text conversion caused by transcription enabling audio capabilities like agent turn detection.<p>SPEV also achieves substantial latency improvements on audio tasks compared to other decision models by skipping the transcription step and operating directly on the raw audio. In our benchmarks SPEV returns an audio native decision in 166 ms compared to 752 ms from OpenAI after including the transcription roundtrip.

  • 情报分类：技术学习与提效
  • 分类依据：内容涉及技术、AI、软件工具或工程实践
  • 信息来源：Hacker News 新项目
  • 发布时间：2026/10/14 23:52:42