Key caveat
A 0–1 output does not prove JAV is “just a classifier.” A sophisticated model can internally perform substantial computation and still expose only a single score. To make that claim confidently, we'd need JAV's architecture and technical documentation.
Is the JAV hype actually worth it?
It’s not the LLM magic we thought it was. It seems to just classify the input into a yes/no output — essentially a score between 0 and 1.
Maybe the real question is: are we overhyping what’s actually a pretty simple classifier?
Use case
JAV: Useful when you need a decision, score, ranking, or classification.
LLM: Useful for generation, summarization, coding, conversation, translation, and many other language tasks.
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
AI has made programming boring.
We used to spend a week building something and feel genuinely proud when it finally worked.
Now you prompt an AI, wait 30 seconds, and it’s done.
The same thing is happening to art, design, music, writing, engineering, 3D, marketing...
Is your website ready for the new iPhone Duo? 👀
Test it in seconds with https://t.co/c21xPJdZrh
→ Built-in proxy. No more “connection refused” nightmares 😭
→ iPhone Duo + 30+ other devices
Drop this in your dev group chat. Your frontend friends will thank you