Electricity Bench

Gemini 3.8 Flash vs Claude Sonnet 5.5

Both grade C- overall. Claude Sonnet 5.5 is about 20x cheaper per task. By suite, real-world issues is even at C- and Claude Sonnet 5.5 takes spec planning (C+ to D+). Both agents run the same private suite cut headless through their own tools, each on its own subscription.Measured

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.

Verdict: Claude Sonnet 5.5 vs Gemini 3.8 Flash

Both grade C- overall. By suite: real-world issues is even at C- (60% vs 60% capability); Claude Sonnet 5.5 takes spec planning, C+ to D+ (74% vs 50% capability); Gemini 3.8 Flash takes vibe coding, C- to D+ (56% vs 48% capability). Gemini 3.8 Flash used 23.7% of Google AI Pro's weekly limit per pass, $0.057 per task at the listed price. Claude Sonnet 5.5 used <1% of Claude Max 5x's weekly limit per pass, $0.003 per task at the listed price. Per task, Claude Sonnet 5.5 is about 20x cheaper ($0.003 vs $0.057). Claude Sonnet 5.5 is faster on real-world issues: a median task takes 37s against 5m 21s. Both finished every task on it.

Changed since W39: Gemini 3.8 Flash: real-world issues F to C-, real-world issues usage 18.0% to 21.9% of the weekly limit, spec planning usage 2.0% to 1.8% of the weekly limit, vibe coding D+ to C-, vibe coding usage 1.2% to 1.1% of the weekly limit.

Gemini 3.8 FlashClears tier 4 (partial tier 5) at 21.9% of Google AI Pro per runC-OVERALLClaude Sonnet 5.5Clears tier 4 (partial tier 5) at 4.0% of Claude Max 5x per runC-OVERALL
real-world@v3C-vsC-
DimensionGemini 3.8 FlashClaude Sonnet 5.5Advantage
Capability60%60%Even
Median task time5m 21s37s4m 44s faster (Claude Sonnet 5.5 leads)
Approx. task cost$0.067 / task$0.002 / task$0.065 / task lower (Claude Sonnet 5.5 leads)

Gemini 3.8 Flash: 21.9% of Google AI Pro (131.3% / 5h, 21.9% / 1w) per run. Its weekly limit holds 6 of the week's 33.6 5h windows. Claude Sonnet 5.5: 4.0% of Claude Max 5x (4.0% / 5h) per run. Its weekly limit holds about 13.8 of the week's 33.6 5h windows. Each figure is a share of that agent's own plan, so there is no delta to draw.

spec-planning@v1D+vsC+
DimensionGemini 3.8 FlashClaude Sonnet 5.5Advantage
Capability50%74%+24% (Claude Sonnet 5.5 leads)
Median task time3m 0s3m 25s24s faster (Gemini 3.8 Flash leads)
Approx. task cost$0.021 / task$0.007 / task$0.014 / task lower (Claude Sonnet 5.5 leads)

Gemini 3.8 Flash: 1.8% of Google AI Pro (11.0% / 5h, 1.8% / 1w) per run. Its weekly limit holds 6 of the week's 33.6 5h windows. Claude Sonnet 5.5: 4.0% of Claude Max 5x (4.0% / 5h) per run. Its weekly limit holds about 13.8 of the week's 33.6 5h windows. Each figure is a share of that agent's own plan, so there is no delta to draw.

vibe-coding@v1C-vsD+
DimensionGemini 3.8 FlashClaude Sonnet 5.5Advantage
Capability56%48%+8% (Gemini 3.8 Flash leads)
Median task time7m 51s56s6m 55s faster (Claude Sonnet 5.5 leads)
Approx. task cost$0.048 / task$0.001 / task$0.047 / task lower (Claude Sonnet 5.5 leads)

Gemini 3.8 Flash: 1.1% of Google AI Pro (6.3% / 5h, 1.1% / 1w) per good solve. Its weekly limit holds 6 of the week's 33.6 5h windows. Claude Sonnet 5.5: 0.2% of Claude Max 5x (0.2% / 5h) per good solve. Its weekly limit holds about 13.8 of the week's 33.6 5h windows. 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 3.8 Flash or Claude Sonnet 5.5?

They are even on our ladder: both clear tier 4 of 5 at 60% capability on real-world@v3, our private suite of real-world coding problems. Latest measurement 2026-09-27.

Which is cheaper per task, Gemini 3.8 Flash or Claude Sonnet 5.5?

Gemini 3.8 Flash used 23.7% of Google AI Pro's weekly limit per pass, $0.057 per task at the listed price. Claude Sonnet 5.5 used <1% of Claude Max 5x's weekly limit per pass, $0.003 per task at the listed price. Per task, Claude Sonnet 5.5 is about 20x cheaper ($0.003 vs $0.057).