Claude Sonnet 5 vs GPT-5.6 Terra: Benchmark Comparison
Claude Sonnet 5 has the higher score on 7 of 22 shared benchmarks; GPT-5.6 Terra leads on 15.
The largest observed score gap is 42.61 pts on IOI , where GPT-5.6 Terra leads.
Reported ±1 standard-error ranges overlap on 10 of 22 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-5.6 Terra | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 41.83% ±3.43 | 41.35% ±3.42 | 0.48 pts | Reported ±1 SE ranges overlap |
| LegalBench | 83.92% ±0.46 | 85.11% ±0.45 | 1.18 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 70.03% ±0.90 | 67.33% ±0.93 | 2.70 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 29.37% ±3.25 | 30.04% ±3.31 | 0.66 pts | Reported ±1 SE ranges overlap |
| TaxEval v2 | 75.63% ±0.84 | 76.17% ±0.84 | 0.53 pts | Reported ±1 SE ranges overlap |
| MedCode | 47.54% ±2.27 | 43.41% ±2.17 | 4.13 pts | Reported ±1 SE ranges overlap |
| MedScribe | 76.05% ±3.05 | 82.87% ±1.95 | 6.81 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 77.00% ±4.23 | 74.00% ±4.41 | 3.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench-Science | 5.71% ±2.79 | 10.00% ±3.61 | 4.29 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 88.89% ±2.22 | 90.91% ±1.91 | 2.02 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 87.55% ±0.37 | 86.66% ±0.33 | 0.89 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 83.01% ±0.90 | 86.47% ±0.82 | 3.47 pts | Reported ±1 SE ranges do not overlap |
| IOI | 45.00% ±2.75 | 87.61% ±6.53 | 42.61 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 82.43% ±1.09 | 85.93% ±1.02 | 3.50 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 | 58.90% ±4.47 | 12.42 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench | 79.60% ±1.80 | 95.40% ±0.94 | 15.80 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 10.61% ±1.51 | 22.73% ±2.31 | 12.12 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 81.33% ±3.05 | 74.59% ±4.20 | 6.73 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 61.91% ±5.74 | 72.08% ±5.41 | 10.18 pts | Reported ±1 SE ranges overlap |
| SRE Bench | 0.38% ±0.38 | 9.54% ±1.82 | 9.16 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 66.03% ±1.23 | 62.38% ±1.26 | 3.65 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Sonnet 5 average | GPT-5.6 Terra average |
|---|---|---|
| Legal | 62.87% | 63.23% |
| Finance | 58.35% | 57.84% |
| Healthcare | 61.80% | 63.14% |
| Math | 77.00% | 74.00% |
| Science | 5.71% | 10.00% |
| Academic | 86.48% | 88.01% |
| Coding | 49.35% | 60.81% |
| Cyber | 31.14% | 40.81% |
| Social Mobility | 66.03% | 62.38% |
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-5.6 Terra cost | Claude Sonnet 5 latency | GPT-5.6 Terra latency |
|---|---|---|---|---|
| Legal Research Bench | $2.72 | $7.91 | 25m46s | 1h10m |
| LegalBench | N/A | N/A | 4.90s | 2.78s |
| MortgageTax | N/A | N/A | 28.67s | 6.32s |
| Tax Agent Bench | $1.79 | $4.83 | 19m19s | 1h01m |
| TaxEval v2 | N/A | N/A | 3m22s | 37.28s |
| MedCode | N/A | N/A | 2m15s | 18.41s |
| MedScribe | N/A | N/A | 4m12s | 35.53s |
| ProofBench v1.1 | $1.37 | $1.27 | 14m58s | 12m57s |
| Terminal-Bench-Science | $30.30 | $5.19 | 3h06m | 2h04m |
| GPQA Diamond | N/A | N/A | 63.23s | 36.76s |
| MMLU Pro | N/A | N/A | 25.19s | 8.17s |
| MMMU Pro | N/A | N/A | 18.82s | 13.03s |
| IOI | $12.89 | $8.66 | 59m44s | 1h23m |
| LiveCodeBench | N/A | N/A | 77.02s | 44.33s |
| ProgramBench | $24.36 | $5.38 | 1h31m | 33m36s |
| SkillsBench | $2.90 | $1.75 | 15m12s | 6m55s |
| SWE-bench | $1.49 | $0.40 | 16m02s | 3m00s |
| Terminal-Bench 4.0 | $26.05 | $5.60 | 1h45m | 31m29s |
| Vibe Code Bench v1.1 | $25.39 | $7.89 | 1h07m | 19m12s |
| CyberBench v1.1 | $1.89 | $3.31 | 16m57s | 17m04s |
| SRE Bench | $18.49 | $30.86 | 3h53m | 3h17m |
| Public Benefits Bench v1.1 | $1.29 | $1.20 | 21m09s | 19m31s |
Results available only for Claude Sonnet 5
None.
Results available only for GPT-5.6 Terra
- SAGE
- Financial Coaching Bench