Claude Opus 4.8 vs GPT-6 Sol: Benchmark Comparison
Claude Opus 4.8 has the higher score on 8 of 16 shared benchmarks; GPT-6 Sol leads on 8.
The largest observed score gap is 25.71 pts on Terminal-Bench Science , where GPT-6 Sol leads.
Reported ±1 standard-error ranges overlap on 4 of 15 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 4.8 | GPT-6 Sol | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 55.10% ±1.01 | 57.54% ±1.01 | 2.44 pts | Reported ±1 SE ranges do not overlap |
| Vals RSI Index | 23.98% | 28.11% | 4.13 pts | Uncertainty comparison unavailable |
| Harvey's Legal Agent Benchmark | 9.58% ±2.15 | 1.67% ±0.82 | 7.92 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 43.75% ±3.45 | 28.85% ±3.15 | 14.90 pts | Reported ±1 SE ranges do not overlap |
| EMB | 69.37% ±2.61 | 71.53% ±2.35 | 2.16 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 53.92% ±0.16 | 49.05% ±0.58 | 4.87 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 64.91% ±3.17 | 53.05% ±3.29 | 11.86 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 53.22% ±2.17 | 47.07% ±2.12 | 6.14 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 85.75% ±1.93 | 82.03% ±1.94 | 3.72 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench Science | 4.29% ±2.44 | 30.00% ±5.52 | 25.71 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 54.79% ±3.35 | 44.79% ±3.36 | 9.99 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 47.25% ±4.18 | 57.20% ±4.21 | 9.95 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 1.00% ±0.70 | 2.00% ±0.99 | 1.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 23.23% ±1.34 | 44.44% ±3.54 | 21.21 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 82.72% ±3.08 | 87.82% ±2.53 | 5.10 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 68.13% ±1.21 | 56.63% ±1.29 | 11.50 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Opus 4.8 average | GPT-6 Sol average |
|---|---|---|
| Legal | 26.67% | 15.26% |
| Finance | 62.73% | 57.88% |
| Healthcare | 69.49% | 64.55% |
| Science | 4.29% | 30.00% |
| Education | 54.79% | 44.79% |
| Coding | 38.55% | 47.87% |
| Social Mobility | 68.13% | 56.63% |
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 4.8 cost | GPT-6 Sol cost | Claude Opus 4.8 latency | GPT-6 Sol latency |
|---|---|---|---|---|
| Vals Index | $13.14 | $7.58 | 39m48s | 29m11s |
| Vals RSI Index | $465.25 | $665.11 | 90h00m | 90h00m |
| Harvey's Legal Agent Benchmark | $10.22 | $3.36 | 24m08s | 13m40s |
| Legal Research Bench | $2.82 | $4.90 | 11m54s | 24m09s |
| EMB | $12.06 | $1.28 | 43m52s | 10m22s |
| Finance Agent (v2) | $4.22 | $2.12 | 9m07s | 9m00s |
| Tax Agent Bench | $1.69 | $2.28 | 7m59s | 17m32s |
| MedCode | N/A | N/A | 105.92s | 62.71s |
| MedScribe | N/A | N/A | 82.36s | 81.14s |
| Terminal-Bench Science | $23.14 | $5.82 | 1h55m | 1h22m |
| SAGE | N/A | N/A | 3m06s | 31.42s |
| Code Migration | $30.51 | $15.68 | 1h15m | 1h26m |
| ProgramBench | $31.27 | $11.67 | 1h32m | 46m11s |
| Terminal-Bench 4.0 | $17.14 | $5.79 | 1h11m | 34m56s |
| Vibe Code Bench v1.1 | $26.88 | $26.36 | 1h16m | 38m56s |
| Public Benefits Bench v1.1 | $1.89 | $2.49 | 12m48s | 31m02s |
Results available only for Claude Opus 4.8
- LegalBench
- MortgageTax
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- MMMU Pro
- LiveCodeBench
- SkillsBench
- SWE-bench
- Time Horizon Index: KSP
Results available only for GPT-6 Sol
- ProofBench v1.1
- BioMysteryBench
- MysteryMechanism
- IOI
- CyberBench v1.1