Open-weight release
Vinci Cyber 30B 1.0
An open-weight defensive infrastructure-as-code specialization of IBM Granite 4.1 30B by SimpleDirect®.
Model cards and weights on Hugging Face
What it is for
Given a configuration and a scanner finding, propose a reviewable change or identify when no edit is appropriate. This is a narrow defensive specialization, not a general vulnerability scanner or an autonomous remediation system.
Repair, no change, recovery
- Repair: propose a bounded edit that addresses the reported finding.
- No change: leave a configuration alone when an edit is not warranted.
- Recovery: reconsider an unsuccessful attempt rather than repeatedly applying a failing change.
These are intended behaviours, not guarantees. Independently validate configuration syntax, security properties and regressions, then obtain human approval before applying a change.
Scope and limitations
This model can change a clean configuration and introduce a security finding, as observed in its internal controls. PatchEval-Verified has not been measured on this checkpoint, and CyberSOCEval is pending. Internal repair scoring is not independently qualified; broad security superiority and safe autonomous operation are not established.
Internal repair scoring remains under qualification and rescoring is pending. Scanner clearance alone does not prove a valid security repair. Results from one model or checkpoint must not be transferred to its sibling or to quantized formats.
Visit the Hugging Face model card for weights, loading guidance and the detailed evaluation record. Evaluate the model in your own environment; this page does not offer hosted inference.
Compare the Cyber model descriptions