Claude Opus 4.7 vs Gemini 3.6 Flash: Benchmark Comparison
Claude Opus 4.7 has the higher score on 12 of 19 shared benchmarks; Gemini 3.6 Flash leads on 6.
The largest observed score gap is 13.97 pts on Tax Agent Bench , 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 3.6 Flash | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 6.67% ±2.00 | 3.33% ±1.17 | 3.33 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 38.46% ±3.38 | 25.00% ±3.01 | 13.46 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 85.25% ±0.45 | 86.70% ±0.41 | 1.45 pts | Reported ±1 SE ranges do not overlap |
| EMB | 63.73% ±3.11 | 65.41% ±2.65 | 1.68 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 51.51% ±0.49 | 56.30% ±0.18 | 4.79 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 70.27% ±0.90 | 67.61% ±0.92 | 2.66 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 63.86% ±3.14 | 49.90% ±3.30 | 13.97 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 75.27% ±0.84 | 74.86% ±0.85 | 0.41 pts | Reported ±1 SE ranges overlap |
| MedCode | 54.86% ±2.21 | 53.15% ±2.16 | 1.70 pts | Reported ±1 SE ranges overlap |
| MedScribe | 82.95% ±1.98 | 79.66% ±1.86 | 3.29 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 90.15% ±1.54 | 93.43% ±1.33 | 3.28 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 89.87% ±0.30 | 89.28% ±0.30 | 0.59 pts | Reported ±1 SE ranges overlap |
| MMMU Pro | 85.55% ±0.84 | 88.38% ±0.77 | 2.83 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 56.10% ±3.37 | 48.69% ±3.38 | 7.42 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 43.88% ±4.22 | 30.93% ±4.07 | 12.95 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 85.07% ±1.01 | 88.08% ±0.94 | 3.01 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 | 79.60% ±1.80 | 2.40 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 71.00% ±4.51 | 64.00% ±4.29 | 7.00 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Claude Opus 4.7 average | Gemini 3.6 Flash average |
|---|---|---|
| Legal | 43.46% | 38.35% |
| Finance | 64.93% | 62.81% |
| Healthcare | 68.91% | 66.41% |
| Academic | 88.52% | 90.36% |
| Education | 56.10% | 48.69% |
| Coding | 56.39% | 52.52% |
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 3.6 Flash cost | Claude Opus 4.7 latency | Gemini 3.6 Flash latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $10.81 | $1.90 | 24m44s | 16m24s |
| Legal Research Bench | $3.06 | $0.47 | 8m32s | 2m48s |
| LegalBench | N/A | N/A | 26.62s | 2.96s |
| EMB | $12.76 | $3.89 | 38m36s | 12m06s |
| Finance Agent (v2) | $4.03 | $1.40 | 6m00s | 3m45s |
| MortgageTax | N/A | N/A | 16.10s | 9.24s |
| Tax Agent Bench | $1.27 | $0.36 | 4m39s | 105.32s |
| TaxEval v2 | N/A | N/A | 62.07s | 12.34s |
| MedCode | N/A | N/A | 54.25s | 16.68s |
| MedScribe | N/A | N/A | 67.50s | 33.55s |
| GPQA Diamond | N/A | N/A | 3m23s | 18.06s |
| MMLU Pro | N/A | N/A | 3m08s | 7.30s |
| MMMU Pro | N/A | N/A | 70.80s | 11.31s |
| SAGE | N/A | N/A | 2m05s | 21.96s |
| Code Migration | $36.39 | $8.93 | 1h12m | 1h30m |
| LiveCodeBench | N/A | N/A | 7m42s | 31.88s |
| ProgramBench | N/A | N/A | 54m07s | 47m33s |
| SWE-bench | $2.42 | $1.19 | 7m22s | 13m08s |
| Vibe Code Bench v1.1 | $21.41 | $3.04 | 35m42s | 25m23s |
Results available only for Claude Opus 4.7
- Vals RSI Index
Results available only for Gemini 3.6 Flash
- BioMysteryBench
- Terminal-Bench Science
- IOI
- CyberBench v1.1
- Public Benefits Bench v1.1