AI Research for Security Operations.

Simbian Research is the research arm of Simbian — publishing benchmarks, models, and tools that advance what AI can do in real threat environments.

Research Publications

[Cyber Defense Benchmark

Agentic threat hunting benchmark. An AI agent receives a threat briefing and 134K raw security events. It must find the attacker by issuing SQL queries and submit exact malicious timestamps. Deterministic scoring.

25 models evaluated

1,080 benchmark runs

$1,267 total API cost

0 models passed

View Benchmark →](/content/research/cyber-defense-benchmark/index.html) [AI SOC LLM Leaderboard

Comprehensive evaluation of frontier models across SOC tasks: alert triage, threat intelligence, log analysis, and incident response. The definitive reference for teams choosing a model for security operations.

View Leaderboard →](/content/best-ai-for-cybersecurity/index.html)