AI Briefing
Sign in

Cyber-OSINT MoE Model Released for Local Threat Intelligence Analysis

·2026.09.29 09:00

Key point

The new Cyber-OSINT model features 26B total parameters with 4B active, a 262K context window, and is trained on 6,500 OSINT/CTI instructions.

Details

A new Cyber-OSINT model has been released, designed for local execution and specialized in cyber threat intelligence (CTI) and open-source intelligence (OSINT) tasks.

Model Architecture and Training

The model is a Mixture of Experts (MoE) architecture with 26B total parameters and 4B active parameters. It supports a 262K context window and was trained on 6,500 OSINT/CTI instructions using Supervised Fine-Tuning (SFT).

Key Capabilities

The model is optimized for investigative work, including:

  • Threat-actor attribution
  • IoC (Indicator of Compromise) pivoting
  • Geolocation
  • Admiralty source grading
moelocal-llmopen-source-modelthreat-intelligenceosintcyber-security
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

KakaoKakao Improves Valid Play Rate by 20% with MoE-Based Personalized Ranking Model Using GDCNRedditIQuestLab Releases IQuest-Q1, a 320B Parameter MoE Model for Agentic CodingRedditWebBrain releases open-source browser agent with 450M local vision modelPyTorchKRHugging Face Adds Native GGUF Execution to TransformersPyTorchKRConvai Unveils 'Laya', a Decision Model Without Text GenerationHacker NewsOllaya Launches Local Runtime for Open-Source Decision Models with TypeSafe SDK Compatibility