Electricity Bench

claude-code + claude-opus-5 vs claude-code + claude-sonnet-5

Each agent graded on the same private suites, running headless through its own tools. Deltas are oriented so a win is a win, whichever way the metric runs.

claude-cliclaude-code + claude-opus-5Solves up to tier 5 at 64.0% of Claude Max 5x per run · 1.6 runs/5hoverall pendingclaude-cliclaude-code + claude-sonnet-5Solves up to tier 5overall pending
real-world@v3CvsF
Dimensionclaude-code + claude-opus-5claude-code + claude-sonnet-5Delta
Capability60%33%27% · claude-code + claude-opus-5
Median latency222605 ms228312 ms5707 ms · claude-code + claude-opus-5
Reliability100%100%even
Error rate0%0%even

claude-code + claude-opus-5: 64.0% of Claude Max 5x per run, about 1.6 runs per 5h. claude-code + claude-sonnet-5: plan cost not measured. Each figure is a share of that agent's own plan, so there is no delta to draw.

Grades and deltas are comparable only within a suite version. See methodology.