[Submitted on 16 Jul 2026 (v1), last revised 27 Jul 2026 (this version, v3)]

View PDF HTML (experimental)

Abstract:Security-agent evaluations commonly measure peak offensive capability under generous inference budgets, emphasizing vulnerability discovery, exploit development, penetration testing, and CTF completion. Such measurements are useful but incomplete: in operational security, every reasoning step, tool call, telemetry query, and enrichment request consumes budget. We evaluate language-model security agents through this cost-success lens on offensive Cybench challenges and defensive Splunk BOTS v1 investigation challenges. Instead of reporting only best-case success, we compare models at fixed cost levels and decompose performance by inference spend and tool spend. Our results show distinct scalingregimes for red- and blue-team tasks. Offensive CTF performance improves with additional test-time compute, and scaled open-weight models can approach frontier proprietary systems while remaining cost-competitive. Defensive SOC investigation does not scale in the same way: success depends more heavily on disciplined tool use, telemetry navigation, and selective enrichment than on raw reasoning budget alone. We argue that security-agent benchmarks should measure economic efficiency and operational fit alongside task success. Cost-aware, SOC-native evaluations provide a clearer picture of which models are practically useful today and where defensive agents still need to improve. We present an interactive website with our results this https URL.

Submission history

From: Paul Kassianik [view email]
[v1] Thu, 16 Jul 2026 17:54:47 UTC (1,025 KB)
[v2] Fri, 17 Jul 2026 17:56:35 UTC (1,052 KB)
[v3] Mon, 27 Jul 2026 17:42:56 UTC (1,230 KB)