Grok 4.6 vs GPT-6 Luna: Benchmark Comparison
Grok 4.6 has the higher score on 12 of 16 shared benchmarks; GPT-6 Luna leads on 4.
The largest observed score gap is 17.79 pts on Legal Research Bench , where Grok 4.6 leads.
Reported ±1 standard-error ranges overlap on 6 of 16 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 | Grok 4.6 | GPT-6 Luna | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Science Index | 38.09% ±1.64 | 28.38% ±1.48 | 9.71 pts | Reported ±1 SE ranges do not overlap |
| Immigration | 65.98% ±1.53 | 62.53% ±1.57 | 3.45 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 48.08% ±3.47 | 30.29% ±3.19 | 17.79 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 33.66% ±3.41 | 19.95% ±2.87 | 13.71 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 44.71% ±2.26 | 44.69% ±2.30 | 0.03 pts | Reported ±1 SE ranges overlap |
| MedScribe | 86.53% ±1.96 | 83.71% ±1.95 | 2.82 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 51.00% ±5.02 | 64.00% ±4.82 | 13.00 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 72.22% ±0.00 | 61.48% ±2.59 | 10.74 pts | Reported ±1 SE ranges do not overlap |
| MysteryMechanism | 30.63% ±3.10 | 19.37% ±2.66 | 11.26 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench-Science | 11.43% ±3.83 | 4.29% ±2.44 | 7.14 pts | Reported ±1 SE ranges do not overlap |
| IOI | 47.61% ±2.09 | 55.56% ±8.87 | 7.95 pts | Reported ±1 SE ranges overlap |
| ProgramBench | 1.00% ±0.70 | 0.50% ±0.50 | 0.50 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 17.17% ±1.34 | 13.64% ±1.51 | 3.54 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 76.24% ±3.82 | 81.65% ±3.38 | 5.41 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 67.68% ±5.87 | 76.25% ±5.29 | 8.57 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 66.85% ±1.23 | 57.65% ±1.28 | 9.20 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Grok 4.6 average | GPT-6 Luna average |
|---|---|---|
| Legal | 57.03% | 46.41% |
| Finance | 33.66% | 19.95% |
| Healthcare | 65.62% | 64.20% |
| Math | 51.00% | 64.00% |
| Science | 38.09% | 28.38% |
| Coding | 35.51% | 37.84% |
| Cyber | 67.68% | 76.25% |
| Social Mobility | 66.85% | 57.65% |
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 | Grok 4.6 cost | GPT-6 Luna cost | Grok 4.6 latency | GPT-6 Luna latency |
|---|---|---|---|---|
| Vals Science Index | $2.45 | $0.11 | 40m10s | 30m15s |
| Immigration | $0.10 | $0.01 | 3m40s | 2m31s |
| Legal Research Bench | $1.53 | $0.44 | 25m04s | 42m28s |
| Tax Agent Bench | $0.98 | $0.19 | 10m58s | 33m16s |
| MedCode | N/A | N/A | 2m07s | 108.77s |
| MedScribe | N/A | N/A | 70.94s | 2m50s |
| ProofBench v1.1 | $0.82 | $0.04 | 14m22s | 7m31s |
| BioMysteryBench | $1.23 | $0.05 | 14m18s | 8m44s |
| MysteryMechanism | $1.31 | $0.03 | 27m55s | 6m01s |
| Terminal-Bench-Science | $4.82 | $0.24 | 1h18m | 1h16m |
| IOI | $8.18 | $0.20 | 3h57m | 35m47s |
| ProgramBench | $16.76 | $0.18 | 4h27m | 28m55s |
| Terminal-Bench 4.0 | $5.06 | $0.35 | 29m25s | 32m46s |
| Vibe Code Bench v1.1 | $4.88 | $1.35 | 25m30s | 37m00s |
| CyberBench v1.1 | $4.07 | $0.13 | 36m52s | 15m50s |
| Public Benefits Bench v1.1 | $0.92 | $0.22 | 25m53s | 52m59s |
Results available only for Grok 4.6
- Vals RSI Index
- LegalBench
- MortgageTax
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- LiveCodeBench
- SkillsBench
- SWE-bench
- Access to Justice Benchmark
- Time Horizon Index: KSP
Results available only for GPT-6 Luna
- Vals Index
- Vals Finance Index
- Vals Legal Index
- HLAB-Vals-1.1
- ClaimBench
- EMB v1.1
- Finance Agent v2.1
- SAGE
- Code Migration v1.1
- SWE-bench Multimodal
- SRE Bench
- Financial Coaching Bench