ZONFORMAT

hey PH! 👋

i’m roni, full-stack dev from india, ex-gsoc @ internet archive.

built ZON because i was tired of json eating 40-70% of my openai bills for no reason.

ZON = zero-overhead notation → same data, 35-70% fewer tokens, zero parse overhead.

works with langchain, llama.cpp, openai sdk, streaming, zod schemas… all in <2kb.

live demo in the first link → paste any json and watch the tokens melt.

would love your brutal feedback (and a star on github if it saves you $ 😂)

https://github.com/ZON-Format

let’s make llm prompts cheap again ⚡️