[KongchangAI]
ModelDolly 2

Dolly 2 Security

A more capable uncensored security model released by Zro, built on GLM-5.3 with abliteration applied to remove alignment. It is intended for defensive testing and red-team workflows.

Timeline (last 90 days)

Oct 28

Zro released two uncensored model variants, Dolly 1 Security and Dolly 2 Security, based on GLM-5.3 Flash and GLM-5.3 respectively for abliteration processing.

Unverified50%
Oct 28

The capability difference between Dolly 1 and Dolly 2 essentially comes from the difference in base specifications, not from differences in Zro's own training investment.

Unverified50%

All Facts (2)

Source Articles