My new Offensive Security model, BaronLLM v3, is ready! 🔥 In this version, I used…
This is a dev post classified by AGK Intelligence as Security (a launch), kept by the Dev Radar because it carries real work, not commentary.
My new Offensive Security model, BaronLLM v3, is ready! 🔥 In this version, I used NVIDIA-Nemotron-3.5-Bolt-E4B as the base model. After fine-tuning it for cybersecurity, I achieved excellent results with partial abliteration. I managed to reduce the refusal rate to 0%, with only a slight loss in thinking and rubric quality caused by the abliteration process. I'll soon share it on Hugging Face, via invitation, with professional cybersecurity experts. 🔥
Posted by Alican Kiraz (39.6k followers) 1 h ago · 24 likes · 1.7k views · view the original post on X. Kept by the Dev Radar as Security.
More dev work like this
- Vulnerabilities are easy to count. Meaningful risk is harder to measure. — @RedHat
- Join us tomorrow for a new episode of Builders Room! — @turnkeyhq
- 🔐 KCSA — Kubernetes and Cloud Native Security Associate — @1ech1ech
- Origin recently shared how it approaches protocol security. Here’s what that looks like… — @OriginProtocol
- The @plether_fi audit competition is live and will run until November 16! — @immunefi
- Shift your open source security from reactive to ready. — @RedHat
- Agents are already running in your environment. Some you approved. Some you have never… — @jetstream_sec
- Attackers are using AI to find #CICD flaws. Defenders can too. — @jfrog
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 31.4k posts from 5k X accounts over the last 21 days, 9.3k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-04 21:21 UTC. Full method.