Claude Opus 4.7 vs Gemini 3.5 Flash: Benchmark Comparison
Claude Opus 4.7 has the higher score on 13 of 19 shared benchmarks; Gemini 3.5 Flash leads on 5.
The largest observed score gap is 22.32 pts on Vibe Code Bench v1.1 , where Claude Opus 4.7 leads.
Reported ±1 standard-error ranges overlap on 8 of 19 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 | Claude Opus 4.7 | Gemini 3.5 Flash | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 6.67% ±2.00 | 2.50% ±0.83 | 4.17 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 38.46% ±3.38 | 30.77% ±3.21 | 7.69 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 85.25% ±0.45 | 83.60% ±0.85 | 1.65 pts | Reported ±1 SE ranges do not overlap |
| EMB | 63.73% ±3.11 | 63.55% ±2.75 | 0.18 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 51.51% ±0.49 | 57.86% ±0.23 | 6.35 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 70.27% ±0.90 | 68.12% ±0.91 | 2.15 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 63.86% ±3.14 | 53.53% ±3.30 | 10.34 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 75.27% ±0.84 | 74.37% ±0.85 | 0.90 pts | Reported ±1 SE ranges overlap |
| MedCode | 54.86% ±2.21 | 55.83% ±2.11 | 0.97 pts | Reported ±1 SE ranges overlap |
| MedScribe | 82.95% ±1.98 | 76.57% ±1.92 | 6.38 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 90.15% ±1.54 | 92.68% ±1.46 | 2.52 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 89.87% ±0.30 | 89.52% ±0.31 | 0.36 pts | Reported ±1 SE ranges overlap |
| MMMU Pro | 85.55% ±0.84 | 88.27% ±0.77 | 2.72 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 56.10% ±3.37 | 49.88% ±3.44 | 6.22 pts | Reported ±1 SE ranges overlap |
| Code Migration | 43.88% ±4.22 | 26.75% ±4.10 | 17.13 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 85.07% ±1.01 | 87.60% ±0.95 | 2.53 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 82.00% ±1.72 | 78.80% ±1.83 | 3.20 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 71.00% ±4.51 | 48.68% ±4.73 | 22.32 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Opus 4.7 average | Gemini 3.5 Flash average |
|---|---|---|
| Legal | 43.46% | 38.96% |
| Finance | 64.93% | 63.48% |
| Healthcare | 68.91% | 66.20% |
| Academic | 88.52% | 90.15% |
| Education | 56.10% | 49.88% |
| Coding | 56.39% | 48.37% |
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 | Claude Opus 4.7 cost | Gemini 3.5 Flash cost | Claude Opus 4.7 latency | Gemini 3.5 Flash latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $10.81 | $2.12 | 24m44s | 7m58s |
| Legal Research Bench | $3.06 | $1.31 | 8m32s | 4m38s |
| LegalBench | N/A | N/A | 26.62s | 34.58s |
| EMB | $12.76 | $5.17 | 38m36s | 18m44s |
| Finance Agent (v2) | $4.03 | $2.51 | 6m00s | 5m22s |
| MortgageTax | N/A | N/A | 16.10s | 11.43s |
| Tax Agent Bench | $1.27 | $0.86 | 4m39s | 3m53s |
| TaxEval v2 | N/A | N/A | 62.07s | 19.69s |
| MedCode | N/A | N/A | 54.25s | 25.29s |
| MedScribe | N/A | N/A | 67.50s | 57.08s |
| GPQA Diamond | N/A | N/A | 3m23s | 27.66s |
| MMLU Pro | N/A | N/A | 3m08s | 26.26s |
| MMMU Pro | N/A | N/A | 70.80s | 12.23s |
| SAGE | N/A | N/A | 2m05s | 32.59s |
| Code Migration | $36.39 | $5.73 | 1h12m | 19m28s |
| LiveCodeBench | N/A | N/A | 7m42s | 64.37s |
| ProgramBench | N/A | N/A | 54m07s | 18m25s |
| SWE-bench | $2.42 | $0.95 | 7m22s | 4m14s |
| Vibe Code Bench v1.1 | $21.41 | $2.54 | 35m42s | 14m53s |
Results available only for Claude Opus 4.7
- Vals RSI Index
Results available only for Gemini 3.5 Flash
- Vals Index
- ProofBench v1.1
- Terminal-Bench Science
- SkillsBench
- Terminal-Bench 4.0
- Public Benefits Bench v1.1