Gemini 3.5 Flash vs Mercury 2.5: Benchmark Comparison
Gemini 3.5 Flash has the higher score on 11 of 11 shared benchmarks; Mercury 2.5 leads on 0.
The largest observed score gap is 45.00 pts on Vibe Code Bench v1.1 , where Gemini 3.5 Flash leads.
Reported ±1 standard-error ranges overlap on 1 of 11 shared benchmarks with comparable uncertainty data. This is not a pairwise statistical significance test.
Shared benchmark results
Scores and reported standard errors come from the latest visible Vals benchmark version available for both models.
| Benchmark | Gemini 3.5 Flash | Mercury 2.5 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 30.77% ±3.21 | 4.33% ±1.41 | 26.44 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 83.60% ±0.85 | 83.06% ±0.48 | 0.54 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 21.65% ±2.96 | 2.49% ±1.02 | 19.16 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 55.83% ±2.11 | 31.33% ±1.95 | 24.50 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 76.57% ±1.92 | 55.09% ±2.10 | 21.48 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 31.00% ±4.65 | 3.00% ±1.71 | 28.00 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench-Science | 5.71% ±2.79 | 0.00% ±0.00 | 5.71 pts | Reported ±1 SE ranges do not overlap |
| SkillsBench | 52.74% ±4.47 | 18.06% ±3.08 | 34.68 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 6.06% ±1.51 | 0.00% ±0.00 | 6.06 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 48.68% ±4.73 | 3.68% ±1.41 | 45.00 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 59.47% ±1.28 | 45.26% ±1.29 | 14.21 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Gemini 3.5 Flash average | Mercury 2.5 average |
|---|---|---|
| Legal | 57.19% | 43.69% |
| Finance | 21.65% | 2.49% |
| Healthcare | 66.20% | 43.21% |
| Math | 31.00% | 3.00% |
| Science | 5.71% | 0.00% |
| Coding | 35.83% | 7.25% |
| Social Mobility | 59.47% | 45.26% |
Cost and latency
Cost per test appears only where the benchmark reports it for both models. Latency is the measured completion time for that benchmark.
| Benchmark | Gemini 3.5 Flash cost | Mercury 2.5 cost | Gemini 3.5 Flash latency | Mercury 2.5 latency |
|---|---|---|---|---|
| Legal Research Bench | $1.31 | $0.06 | 4m38s | 47.61s |
| LegalBench | N/A | N/A | 34.58s | 3.91s |
| Tax Agent Bench | $0.86 | $0.02 | 3m53s | 29.67s |
| MedCode | N/A | N/A | 25.29s | 7.93s |
| MedScribe | N/A | N/A | 57.08s | 9.99s |
| ProofBench v1.1 | $0.57 | $0.04 | 6m34s | 3m22s |
| Terminal-Bench-Science | $3.80 | $0.18 | 1h01m | 28m14s |
| SkillsBench | $1.25 | $0.10 | 7m15s | 49.36s |
| Terminal-Bench 4.0 | $5.98 | $0.37 | 1h11m | 15m14s |
| Vibe Code Bench v1.1 | $2.54 | $0.38 | 14m53s | 3m13s |
| Public Benefits Bench v1.1 | $1.11 | $0.07 | 5m39s | 68.33s |
Results available only for Gemini 3.5 Flash
- MortgageTax
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- MMMU Pro
- SAGE
- LiveCodeBench
- ProgramBench
- SWE-bench
Results available only for Mercury 2.5
- IOI
- CyberBench v1.1