gradientsmith
← leaderboard

claude-haiku-4-5

200 rollouts

pass@1 clean
78%
pass@1 adversarial
66%
robustness gap
12%
p50 latency
700 ms
p95 latency
1600 ms
cost per attempt
$0.00070
cost per solved clean
$0.0009
cost per solved adv
$0.0011

Per-task breakdown

taskfamilycleanadversarial
code/merge-intervals/001codepasspass
code/max-subarray-sum/001codepasspass
code/roman-to-int/001codepasspass
structured/receipt-cents/001structuredpasspass
structured/apache-log-extract/001structuredpassfail