Jevv Week: Making LLMs Cheap and Fast

Sep 30, 2026 · Lenny's Podcast
🎧 PodShort 46 min squeezed to 3 SaaS HoarderSH New
Episode artwork
Claire Belle
Product Lead and AI Obsessive at How AI
John Linquist
Product Lead and AI Obsessive at How AI
Lenny's Podcast
46 min squeezed to 3
Full episode from Lenny's Podcast
Quotable Moments

The speed and cost-efficiency of Jevv allow for rapid iteration and experimentation with large datasets that would be prohibitively expensive with traditional LLMs.

John LinquistProduct Lead and AI Obsessive at How AI
04:01

Jevv, a new type of model, focuses on structured output rather than free-form text, enabling faster and more cost-effective AI applications.

John LinquistProduct Lead and AI Obsessive at How AI
00:17

This is like a life coach of like, 'you need to make sure you hit all these things, I'm gonna watch everything you do, every single word you say, and make sure that you did the thing you said you would do.'

John LinquistProduct Lead and AI Obsessive at How AI
34:34
Key Insights
  • Jevv, a new type of model, focuses on structured output rather than free-form text, enabling faster and more cost-effective AI applications.
  • Jevv models offer significant cost savings, with one user reporting spending only $0.73 on thousands of demos and large-scale data processing.
  • The speed and cost-efficiency of Jevv allow for rapid iteration and experimentation with large datasets that would be prohibitively expensive with traditional LLMs.
  • Jevv excels at tasks requiring multi-step reasoning and classification, like merging duplicate records in large datasets or analyzing complex text inputs for specific actions.
  • For tasks with well-defined inputs and a limited set of discrete outputs, such as in gaming or UI navigation, Jevv can provide a more efficient and faster solution than broad LLMs.
  • Jevv's strength lies in structured decision-making and routing, making it ideal for scenarios where the AI needs to follow specific rules or execute pre-defined actions.
  • The cost and speed advantages of Jevv are particularly impactful for developers who need to build applications that can handle massive amounts of data or perform complex multi-step reasoning.
  • AI adoption is growing, leading to increased security risks and requirements, and Jevv offers a scalable solution to manage these challenges.
Metrics Mentioned
  • $0.73 spent on thousands of demos and large-scale data processing (User's cost for using Jevv for data processing.)
  • 10x faster average move (Jevv's performance compared to a traditional LLM in a chess game benchmark.)
  • 4x cheaper (Jevv's cost compared to a traditional LLM in a chess game benchmark.)
  • 82% reduction in audit prep time (Claimed efficiency improvement from using Vanta's platform.)
  • $1000 off (A promotional offer for Vanta's platform for podcast listeners.)
  • $4 (approximate cost per million input tokens) (Estimated cost for using Jevv, with no cost for output tokens.)
🎧Full Episode:Lenny's Podcast →