Electricity Bench

claude-sonnet-5 vs gpt-5.6-luna

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

claude-sonnet-5Clears tier 3 (partial tier 5) at 9.7% of Claude Max 20x per runDOVERALLgpt-5.6-lunaClears no tier (partial tier 5) at 3.0% of ChatGPT Plus per runDOVERALL
real-world@v3DvsD
Dimensionclaude-sonnet-5gpt-5.6-lunaAdvantage
Capability53%51%+3% (claude-sonnet-5 leads)
Median task time5m 6s2m 15s2m 51s faster (gpt-5.6-luna leads)
Approx. task cost$0.009 / task$0.009 / task$0.000 / task lower (claude-sonnet-5 leads)

claude-sonnet-5: about 9.7% of Claude Max 20x (about 9.7% / 5h, 1.0% / 1w) per run (estimated: 2 of 15 cells measured no share of their own). gpt-5.6-luna: 3.0% of ChatGPT Plus (3.0% / 1w) per run. Each figure is a share of that agent's own plan, so there is no delta to draw.

vibe-coding@v1DvsC
Dimensionclaude-sonnet-5gpt-5.6-lunaAdvantage
Capability37%56%+19% (gpt-5.6-luna leads)
Median task time5m 21s2m 59s2m 23s faster (gpt-5.6-luna leads)
Approx. task cost$0.005 / task$0.012 / task$0.007 / task lower (claude-sonnet-5 leads)

claude-sonnet-5: 0.4% of Claude Max 20x (0.4% / 5h, 0.2% / 1w) per good solve. gpt-5.6-luna: 0.3% of ChatGPT Plus (0.3% / 1w) per good solve. 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.

Frequently asked questions

Which is better for coding, claude-sonnet-5 or gpt-5.6-luna?

claude-sonnet-5 clears higher: tier 3 of 5 versus tier 0 for gpt-5.6-luna on real-world@v3, our private suite of real-world coding problems. Latest measurement 2026-08-11.

How much do claude-sonnet-5 and gpt-5.6-luna cost?

claude-sonnet-5 runs on Claude Max 20x at $200/month (public price as of 2026-08-05). gpt-5.6-luna runs on ChatGPT Plus at $20/month (public price as of 2026-07-20).