Skip to main content

TasksDropbear Broken Authentication Detection

Binary AnalysisDropbear12% pass rateView TaskView Prompt

Detect backdoor in Dropbear SSH server: password auth handler replaced so send_msg_userauth_failure() calls send_msg_userauth_success(), accepting any password.

Performance

ModelPass RateRunsAvg CostAvg Time
Googlegemini-3-flash-preview67%
$0.405m
OpenAIgpt-5.3-codex-xhigh67%
$1.1520m
Anthropicclaude-opus-4.667%
$7.6289m
Anthropicclaude-sonnet-4.533%
$0.616m
Googlegemini-3-pro-preview33%
$0.858m
OpenAIgpt-5.4-xhigh33%
$1.8470m
Grokgrok-4.1-fast0%
$0.023m
Z.aiglm-50%
$0.0910m
DeepSeekdeepseek-v3.20%
$0.1610m
OpenAIgpt-5.40%
$0.222m
Kimikimi-k2.50%
$0.2210m
Googlegemini-2.5-pro0%
$0.295m
OpenAIgpt-5.20%
$0.3411m
OpenAIgpt-5.3-codex0%
$0.354m
OpenAIgpt-50%
$0.3711m
OpenAIgpt-5.2-codex0%
$0.447m
Grokgrok-40%
$0.4610m
Z.aiglm-4.70%
$0.5017m
Anthropicclaude-sonnet-40%
$0.594m
Anthropicclaude-haiku-4.50%
$0.608m
OpenAIgpt-5.3-codex-high0%
$0.6314m
OpenAIgpt-5.4-high0%
$0.8311m
OpenAIgpt-5.2-codex-high0%
$1.4417m
Googlegemini-3.1-pro-preview0%
$1.6320m
OpenAIgpt-5.2-codex-xhigh0%
$2.0239m
Anthropicclaude-opus-4.50%
$2.5338m