# Sable: a verifiable inference system with cryptographic proofs for model weights

> Original title: Introducing Sable, our verifiable inference system.

- Source: X search: AI launch posts (introducing, just launched)
- Published: 2026-11-03T16:40:03.000Z
- Bad Influence: https://badinfluence.news/items/ptq7gbdtgjq9rrjcttkuvruxb
- Original: https://x.com/zroai_/status/2117658074668282199

## Summary

Sable is a verifiable inference system that produces cryptographic proofs of committed weights, quantization, and sampled randomness. The system will soon be deployed for Dolly models, according to the announcement, with code and a paper available on GitHub.

## Post

Introducing Sable, our verifiable inference system.

Soon to be deployed for our Dolly models - cryptographic proofs of committed weights, quantization, sampled randomness and more! 

Code and paper: https://github.com/moonmath-ai/Sable

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWNr8qNXsAAcFc5.png&mode=thumb&exp=1793923200&sig=7d19d51febf0dd97)