Claude Sonnet 5 vs GPT 5.5: Benchmark Comparison
Claude Sonnet 5 has the higher score on 9 of 21 shared benchmarks; GPT 5.5 leads on 12.
The largest observed score gap is 15.73 pts on SkillsBench , where GPT 5.5 leads.
Reported ±1 standard-error ranges overlap on 12 of 21 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 Sonnet 5 | GPT 5.5 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 5.00% ±1.65 | 3.75% ±1.17 | 1.25 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 41.83% ±3.43 | 40.38% ±3.41 | 1.44 pts | Reported ±1 SE ranges overlap |
| LegalBench | 83.92% ±0.46 | 86.52% ±0.41 | 2.59 pts | Reported ±1 SE ranges do not overlap |
| EMB | 66.32% ±3.01 | 64.54% ±2.87 | 1.77 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 53.91% ±0.52 | 51.76% ±0.55 | 2.15 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 70.03% ±0.90 | 68.76% ±0.91 | 1.27 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 62.27% ±3.19 | 60.46% ±3.22 | 1.82 pts | Reported ±1 SE ranges overlap |
| TaxEval v2 | 75.63% ±0.84 | 74.98% ±0.85 | 0.65 pts | Reported ±1 SE ranges overlap |
| MedCode | 47.54% ±2.27 | 49.10% ±2.19 | 1.56 pts | Reported ±1 SE ranges overlap |
| MedScribe | 76.05% ±3.05 | 86.87% ±1.93 | 10.81 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 88.89% ±2.22 | 93.18% ±1.29 | 4.29 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 87.55% ±0.37 | 88.14% ±0.32 | 0.60 pts | Reported ±1 SE ranges overlap |
| MMMU Pro | 83.01% ±0.90 | 88.27% ±0.77 | 5.26 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 48.92% ±3.40 | 51.53% ±3.95 | 2.61 pts | Reported ±1 SE ranges overlap |
| Code Migration | 44.39% ±4.25 | 45.16% ±4.16 | 0.77 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 82.43% ±1.09 | 85.30% ±1.02 | 2.87 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 0.50% ±0.50 | 0.50 pts | Reported ±1 SE ranges overlap |
| SkillsBench | 46.48% ±4.49 | 62.21% ±4.52 | 15.73 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench | 79.60% ±1.80 | 82.60% ±1.70 | 3.00 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 81.33% ±3.05 | 69.85% ±4.54 | 11.48 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 66.03% ±1.23 | 60.89% ±1.27 | 5.14 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Sonnet 5 average | GPT 5.5 average |
|---|---|---|
| Legal | 43.58% | 43.55% |
| Finance | 65.63% | 64.10% |
| Healthcare | 61.80% | 67.98% |
| Academic | 86.48% | 89.86% |
| Education | 48.92% | 51.53% |
| Coding | 55.70% | 57.60% |
| Social Mobility | 66.03% | 60.89% |
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 Sonnet 5 cost | GPT 5.5 cost | Claude Sonnet 5 latency | GPT 5.5 latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $8.95 | $4.60 | 38m52s | 12m14s |
| Legal Research Bench | $2.72 | $7.40 | 25m46s | 37m45s |
| LegalBench | N/A | N/A | 4.90s | 18.14s |
| EMB | $10.29 | $3.27 | 49m46s | 15m11s |
| Finance Agent (v2) | $0.75 | $4.15 | 13m12s | 11m02s |
| MortgageTax | N/A | N/A | 28.67s | 28.24s |
| Tax Agent Bench | $1.79 | $4.07 | 19m19s | 15m52s |
| TaxEval v2 | N/A | N/A | 3m22s | 95.83s |
| MedCode | N/A | N/A | 2m15s | 2m40s |
| MedScribe | N/A | N/A | 4m12s | 2m13s |
| GPQA Diamond | N/A | N/A | 63.23s | 112.33s |
| MMLU Pro | N/A | N/A | 25.19s | 42.11s |
| MMMU Pro | N/A | N/A | 18.82s | 54.15s |
| SAGE | N/A | N/A | 7m14s | 76.05s |
| Code Migration | $35.31 | $6.44 | 1h57m | 31m35s |
| LiveCodeBench | N/A | N/A | 77.02s | 2m47s |
| ProgramBench | $24.36 | $6.95 | 1h31m | 22m03s |
| SkillsBench | $2.90 | $2.54 | 15m12s | 6m55s |
| SWE-bench | $1.49 | $1.36 | 16m02s | 7m06s |
| Vibe Code Bench v1.1 | $25.39 | $16.66 | 1h07m | 31m52s |
| Public Benefits Bench v1.1 | $1.29 | $3.97 | 21m09s | 39m37s |
Results available only for Claude Sonnet 5
- Vals Index
- ProofBench v1.1
- Terminal-Bench Science
- IOI
- Terminal-Bench 4.0
- Vibe Code Bench 1-100
- CyberBench v1.1
Results available only for GPT 5.5
- Vals RSI Index
- SRE Bench
- Time Horizon Index: KSP