GitHub swaps an LLM for a small ModernBERT model in secret scanning

Name: GitHub swapped an LLM for a tiny model to catch leaked secrets
Uploaded: 2026-10-14T00:00:00+05:30
Description: Regex candidates + fine-tuned ModernBERT classifier: 11.5% higher owner-weighted acceptance, 86% lower median scan time and 98%+ lower serving cost vs the previous LLM pipeline; offline 13% higher precision, 19% higher recall. Now extends generic detection to push protection.

Why it matters

Regex candidates + fine-tuned ModernBERT classifier: 11.5% higher owner-weighted acceptance, 86% lower median scan time and 98%+ lower serving cost vs the previous LLM pipeline; offline 13% higher precision, 19% higher recall. Now extends generic detection to push protection.

Read the source: GitHub Blog (opens in a new tab)

Watch on YouTube Watch on Instagram

All hits from 14 Oct 2026 More Security Get the daily hits by email