List your company

Vulnerability Attack Analysis on WebGenBench Social Media target-domain and cross-domain 1.0

80.65ASR (CodeQL)GPT-5.2
48.77457.049565.32573.6005Feb 2, 2026

Evaluation results

MethodLinks
GPT-5.2Evaluation Protocol=Cross-domain2026.0280.6587.580.6580
GPT-5.2Evaluation Protocol=Held-out (target-domain)2026.0280757052.5
Claude-4.5 OpusEvaluation Protocol=Cross-domain2026.028087.58087.5
Claude-4.5 OpusEvaluation Protocol=Held-out (target-domain)2026.0275807577.5
ComposerEvaluation Protocol=Held-out (target-domain)2026.0275755075
GrokEvaluation Protocol=Held-out (target-domain)2026.02758062.560.61
Gemini-3 ProEvaluation Protocol=Cross-domain2026.027581.257576.47
ComposerEvaluation Protocol=Cross-domain2026.027581.8271.4381.25
GrokEvaluation Protocol=Cross-domain2026.027572.227566.67
Gemini-3 ProEvaluation Protocol=Held-out (target-domain)2026.0260806066.67
Gemini-3 FlashEvaluation Protocol=Cross-domain2026.0257.147057.1457.14
Gemini-3 FlashEvaluation Protocol=Held-out (target-domain)2026.025066.675066.67