Kriptic.near

$NEAR AI just launched uncensored AI models for security teams, plus a program called NEAR AI RED. Here is why that matters. 👇 Think of a boxer getting ready for a big fight. A sparring partner who refuses to throw a punch is useless. And you train behind closed doors, so no rival scout sees your weak side. Security work is the same. Ask a mainstream AI how your code could be hacked, and it often refuses. It can't tell a defender from an attacker. Attackers don't have that problem. Railway says the group that breached it this month likely had uncensored models. So NEAR AI now offers "abliterated" versions of DeepSeek V4.1 Flash and Qwen 3.8. Their built-in refusals are reduced, so they help with legit security testing. They run inside sealed hardware on NEAR AI Cloud. NEAR AI says not even it can see your prompts, and you can check that yourself. The details at launch: • Open to every NEAR AI Cloud account, no application needed • DeepSeek V4.1 Flash: $0.30 in, $1.20 out per 1M tokens • Qwen 3.8 27B: $0.04 in, $1.35 out per 1M tokens NEAR AI RED is optional, for security teams: • 20% off the first month • Early access to new models • Workshops and direct support • Members agree to test only systems they own or have permission to test Try it: near.ai/cybersecurity Defenders finally get a partner that punches back. Behind closed doors. 🥊🔒🛡️⚡️ Learnt something today? Follow for more!