Electricity Bench

claude-code + claude-fable-5 vs codex-cli + gpt-5.6-luna

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-fable-5Solves up to tier 5overall pendingcodex-clicodex-cli + gpt-5.6-lunaSolves up to tier 5 at 2.8% of ChatGPT Plus per run · 35.8 runs/weekoverall pending
real-world@v3CvsD
Dimensionclaude-code + claude-fable-5codex-cli + gpt-5.6-lunaDelta
Capability60%47%13% · claude-code + claude-fable-5
Median latency169381 ms94668 ms74713 ms · codex-cli + gpt-5.6-luna
Reliability100%100%even
Error rate0%0%even

claude-code + claude-fable-5: plan cost not measured. codex-cli + gpt-5.6-luna: 2.8% of ChatGPT Plus per run, about 35.8 runs per week. 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.