Electricity Bench

gemini-3.7-flash vs composer-2.5

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.

gemini-3.7-flashClears tier 4 (partial tier 5) at 9.0% of Google AI Pro per runCOVERALLcomposer-2.5Clears tier 4 (partial tier 5) at 1.9% of Cursor Pro per runCOVERALL
real-world@v3CvsC
Dimensiongemini-3.7-flashcomposer-2.5Advantage
Capability73%60%+13% (gemini-3.7-flash leads)
Median task time2m 25s3m 52s1m 27s faster (gemini-3.7-flash leads)
Approx. task cost$0.028 / task$0.025 / task$0.003 / task lower (composer-2.5 leads)

gemini-3.7-flash: 9.0% of Google AI Pro (54.0% / 5h, 9.0% / 1w) per run. composer-2.5: 1.9% of Cursor Pro (1.9% / 1m) per run. Each figure is a share of that agent's own plan, so there is no delta to draw.

vibe-coding@v1CvsD
Dimensiongemini-3.7-flashcomposer-2.5Advantage
Capability72%52%+20% (gemini-3.7-flash leads)
Median task time3m 37s4m 56s1m 19s faster (gemini-3.7-flash leads)
Approx. task cost$0.031 / task$0.025 / task$0.005 / task lower (composer-2.5 leads)

gemini-3.7-flash: 0.7% of Google AI Pro (4.0% / 5h, 0.7% / 1w) per good solve. composer-2.5: 0.1% of Cursor Pro (0.1% / 1m) 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, gemini-3.7-flash or composer-2.5?

Both clear tier 4 of 5, but gemini-3.7-flash scores higher on capability: 73% versus 60% on real-world@v3, our private suite of real-world coding problems. Latest measurement 2026-08-13.