Muse Spark 1.2 vs GPT-6 Luna: Benchmark Comparison
Muse Spark 1.2 has the higher score on 7 of 18 shared benchmarks; GPT-6 Luna leads on 10.
The largest observed score gap is 33.78 pts on IOI , where GPT-6 Luna leads.
Reported ±1 standard-error ranges overlap on 7 of 18 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 | Muse Spark 1.2 | GPT-6 Luna | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 49.29% ±1.10 | 51.22% ±1.07 | 1.93 pts | Reported ±1 SE ranges overlap |
| Harvey's Legal Agent Benchmark | 25.42% ±3.67 | 2.92% ±1.36 | 22.50 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 43.75% ±3.45 | 30.29% ±3.19 | 13.46 pts | Reported ±1 SE ranges do not overlap |
| EMB | 56.98% ±3.13 | 68.52% ±2.81 | 11.54 pts | Reported ±1 SE ranges do not overlap |
| Finance Agent (v2) | 60.60% ±0.28 | 49.87% ±0.23 | 10.73 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 56.86% ±2.32 | 58.86% ±3.21 | 2.00 pts | Reported ±1 SE ranges overlap |
| MedCode | 49.35% ±2.19 | 44.69% ±2.30 | 4.66 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 90.06% ±1.96 | 83.71% ±1.95 | 6.35 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 43.00% ±4.98 | 64.00% ±4.82 | 21.00 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 64.81% ±1.33 | 61.48% ±2.59 | 3.33 pts | Reported ±1 SE ranges overlap |
| SAGE | 47.66% ±3.44 | 48.09% ±3.38 | 0.43 pts | Reported ±1 SE ranges overlap |
| Code Migration | 29.95% ±4.02 | 42.55% ±4.42 | 12.60 pts | Reported ±1 SE ranges do not overlap |
| IOI | 21.78% ±0.87 | 55.56% ±8.87 | 33.78 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.50% ±0.50 | 0.50% ±0.50 | 0.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 6.06% ±1.51 | 13.64% ±1.51 | 7.57 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 79.10% ±3.31 | 81.65% ±3.38 | 2.55 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 69.52% ±5.56 | 76.25% ±5.29 | 6.73 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 68.47% ±1.21 | 57.65% ±1.28 | 10.83 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Muse Spark 1.2 average | GPT-6 Luna average |
|---|---|---|
| Legal | 34.58% | 16.60% |
| Finance | 58.15% | 59.08% |
| Healthcare | 69.70% | 64.20% |
| Math | 43.00% | 64.00% |
| Science | 64.81% | 61.48% |
| Education | 47.66% | 48.09% |
| Coding | 27.48% | 38.78% |
| Cyber | 69.52% | 76.25% |
| Social Mobility | 68.47% | 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 | Muse Spark 1.2 cost | GPT-6 Luna cost | Muse Spark 1.2 latency | GPT-6 Luna latency |
|---|---|---|---|---|
| Vals Index | $1.94 | $0.43 | 22m56s | 30m34s |
| Harvey's Legal Agent Benchmark | $2.09 | $0.30 | 24m16s | 17m32s |
| Legal Research Bench | $0.51 | $0.44 | 5m07s | 42m28s |
| EMB | $2.31 | $0.14 | 17m07s | 17m21s |
| Finance Agent (v2) | $0.77 | $0.12 | 5m04s | 14m58s |
| Tax Agent Bench | $0.27 | $0.19 | 119.91s | 33m16s |
| MedCode | N/A | N/A | 60.17s | 108.77s |
| MedScribe | N/A | N/A | 61.46s | 2m50s |
| ProofBench v1.1 | $0.44 | $0.04 | 7m49s | 7m31s |
| BioMysteryBench | $0.99 | $0.05 | 23m37s | 8m44s |
| SAGE | N/A | N/A | 29.07s | 91.07s |
| Code Migration | $3.78 | $0.60 | 31m07s | 50m23s |
| IOI | $2.64 | $0.20 | 25m35s | 35m47s |
| ProgramBench | $2.47 | $0.18 | 26m38s | 28m55s |
| Terminal-Bench 4.0 | $4.14 | $0.35 | 1h18m | 32m46s |
| Vibe Code Bench v1.1 | $1.53 | $1.35 | 19m55s | 37m00s |
| CyberBench v1.1 | $2.15 | $0.13 | 20m21s | 15m50s |
| Public Benefits Bench v1.1 | $0.59 | $0.22 | 8m55s | 52m59s |
Results available only for Muse Spark 1.2
- LegalBench
- MortgageTax
- TaxEval v2
- MMLU Pro
- MMMU Pro
- SkillsBench
- SWE-bench
Results available only for GPT-6 Luna
- MysteryMechanism
- Terminal-Bench Science