Ranking / Artificial Analysis Cyber Index

Artificial Analysis Cyber Index

Data updated 13 Oct 2026

Bucket

Supporting evidence

Unit

index

Direction

Higher is better

Version

v1.0

Display harness

Artificial Analysis Stirrup · Cyber Index v1.0 · max

Board

https://artificialanalysis.ai/evaluations/artificial-analysis-cyber-index

Display only. This composite measures defensive source-code tasks. Successful capability scores depend on safety refusals, and refusal rates are reported separately. It does not enter weighted Overall.

Artificial Analysis v1.0, Stirrup. Equal mean of CWE-Bench-AA pass@1, DeepsecBench-AA median F2 over three repeats, and CyberGym-E2E-AA pass@1. DeepsecBench uses a GPT-5.6 Sol High judge. Public-model configurations retain their source effort and fallback labels on the effort and model pages. Trusted-access variants require a distinct catalog identity. Methodology

Compare published benchmark results with category weights →

Models

Results are ordered by score in the display harness, followed by models with no result. Other harnesses are listed separately below. Missing results remain unknown.

26–50 of 469 entries

Other harnesses

These runs use other harnesses. Capability scoring matches configurations separately; model details identify which source records contribute.