GPT-6 Luna vs Hy4 Preview: Benchmark Comparison
GPT-6 Luna has the higher score on 10 of 21 shared benchmarks; Hy4 Preview leads on 11.
The largest observed score gap is 14.90 pts on Legal Research Bench , where Hy4 Preview leads.
Reported ±1 standard-error ranges overlap on 10 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 | GPT-6 Luna | Hy4 Preview | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Finance Index | 53.05% ±1.43 | 57.51% ±1.45 | 4.45 pts | Reported ±1 SE ranges do not overlap |
| Vals Index | 52.32% ±1.12 | 53.21% ±1.13 | 0.89 pts | Reported ±1 SE ranges overlap |
| Vals Legal Index | 62.20% ±1.57 | 71.70% ±1.84 | 9.50 pts | Reported ±1 SE ranges do not overlap |
| HLAB-Vals-1.1 | 4.17% ±1.63 | 8.75% ±2.19 | 4.58 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 30.29% ±3.19 | 45.19% ±3.46 | 14.90 pts | Reported ±1 SE ranges do not overlap |
| EMB v1.1 | 69.95% ±2.54 | 68.95% ±2.40 | 1.01 pts | Reported ±1 SE ranges overlap |
| Finance Agent v2.1 | 56.84% ±2.12 | 61.62% ±2.06 | 4.78 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 19.95% ±2.87 | 32.62% ±3.38 | 12.67 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 44.69% ±2.30 | 43.25% ±2.13 | 1.44 pts | Reported ±1 SE ranges overlap |
| MedScribe | 83.71% ±1.95 | 83.60% ±2.06 | 0.11 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 64.00% ±4.82 | 75.00% ±4.35 | 11.00 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 61.48% ±2.59 | 69.26% ±2.59 | 7.78 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench-Science | 4.29% ±2.44 | 1.43% ±1.43 | 2.86 pts | Reported ±1 SE ranges overlap |
| Code Migration v1.1 | 46.67% ±3.61 | 45.36% ±3.59 | 1.30 pts | Reported ±1 SE ranges overlap |
| IOI | 55.56% ±8.87 | 59.33% ±4.60 | 3.78 pts | Reported ±1 SE ranges overlap |
| ProgramBench | 0.50% ±0.50 | 0.00% ±0.00 | 0.50 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 13.64% ±1.51 | 8.08% ±1.34 | 5.55 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 81.65% ±3.38 | 77.48% ±4.04 | 4.17 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 76.25% ±5.29 | 65.36% ±5.55 | 10.89 pts | Reported ±1 SE ranges do not overlap |
| SRE Bench | 2.67% ±1.00 | 2.29% ±0.93 | 0.38 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 57.65% ±1.28 | 68.61% ±1.21 | 10.96 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | GPT-6 Luna average | Hy4 Preview average |
|---|---|---|
| Legal | 17.23% | 26.97% |
| Finance | 48.91% | 54.40% |
| Healthcare | 64.20% | 63.42% |
| Math | 64.00% | 75.00% |
| Science | 32.88% | 35.34% |
| Coding | 39.60% | 38.05% |
| Cyber | 39.46% | 33.82% |
| Social Mobility | 57.65% | 68.61% |
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 | GPT-6 Luna cost | Hy4 Preview cost | GPT-6 Luna latency | Hy4 Preview latency |
|---|---|---|---|---|
| Vals Finance Index | $0.15 | $0.68 | 22m01s | 32m15s |
| Vals Index | $0.45 | $1.45 | 32m23s | 59m23s |
| Vals Legal Index | $0.31 | $0.81 | 32m30s | 46m04s |
| HLAB-Vals-1.1 | $0.30 | $0.98 | 21m45s | 33m09s |
| Legal Research Bench | $0.44 | $0.85 | 42m28s | 1h04m |
| EMB v1.1 | $0.14 | $0.79 | 17m21s | 30m39s |
| Finance Agent v2.1 | $0.12 | $0.67 | 15m27s | 25m03s |
| Tax Agent Bench | $0.19 | $0.59 | 33m16s | 41m04s |
| MedCode | N/A | N/A | 108.77s | 6m39s |
| MedScribe | N/A | N/A | 2m50s | 5m50s |
| ProofBench v1.1 | $0.04 | $0.34 | 7m31s | 22m54s |
| BioMysteryBench | $0.05 | $0.36 | 8m44s | 23m30s |
| Terminal-Bench-Science | $0.24 | $2.14 | 1h16m | 3h48m |
| Code Migration v1.1 | $0.83 | $3.58 | 1h00m | 1h54m |
| IOI | $0.20 | $1.35 | 35m47s | 1h01m |
| ProgramBench | $0.18 | $12.81 | 28m55s | 2h10m |
| Terminal-Bench 4.0 | $0.35 | $2.10 | 32m46s | 2h10m |
| Vibe Code Bench v1.1 | $1.35 | $2.18 | 37m00s | 41m33s |
| CyberBench v1.1 | $0.13 | $0.49 | 15m50s | 36m43s |
| SRE Bench | $1.55 | $4.90 | 1h21m | 1h40m |
| Public Benefits Bench v1.1 | $0.22 | $0.25 | 52m59s | 1h03m |
Results available only for GPT-6 Luna
- Vals Science Index
- ClaimBench
- MysteryMechanism
- SAGE
- SWE-bench Multimodal
Results available only for Hy4 Preview
- LegalBench