Muse Spark 1.3 vs GPT-5.6 Luna: Benchmark Comparison
Muse Spark 1.3 has the higher score on 8 of 14 shared benchmarks; GPT-5.6 Luna leads on 6.
The largest observed score gap is 21.67 pts on Harvey's Legal Agent Benchmark , where Muse Spark 1.3 leads.
Reported ±1 standard-error ranges overlap on 8 of 14 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 1.3 | GPT-5.6 Luna | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 53.20% ±1.07 | 51.69% ±1.06 | 1.51 pts | Reported ±1 SE ranges overlap |
| Harvey's Legal Agent Benchmark | 22.92% ±3.42 | 1.25% ±0.83 | 21.67 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 40.87% ±3.42 | 36.54% ±3.35 | 4.33 pts | Reported ±1 SE ranges overlap |
| EMB | 62.71% ±2.96 | 67.12% ±2.94 | 4.40 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 58.90% ±0.44 | 55.04% ±0.31 | 3.86 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 71.93% ±2.92 | 60.81% ±3.26 | 11.12 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 55.00% ±5.00 | 60.00% ±4.92 | 5.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench Science | 4.29% ±2.44 | 0.00% ±0.00 | 4.29 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 27.58% ±4.07 | 44.55% ±4.24 | 16.97 pts | Reported ±1 SE ranges do not overlap |
| IOI | 43.94% ±1.45 | 61.78% ±11.61 | 17.83 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.50% ±0.50 | 0.00% ±0.00 | 0.50 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 10.61% ±1.75 | 11.62% ±1.01 | 1.01 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 82.86% ±2.90 | 77.06% ±3.07 | 5.80 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 69.41% ±5.76 | 73.63% ±5.56 | 4.23 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Muse Spark 1.3 average | GPT-5.6 Luna average |
|---|---|---|
| Legal | 31.89% | 18.89% |
| Finance | 64.52% | 60.99% |
| Math | 55.00% | 60.00% |
| Science | 4.29% | 0.00% |
| Coding | 33.10% | 39.00% |
| Cyber | 69.41% | 73.63% |
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 | Muse Spark 1.3 cost | GPT-5.6 Luna cost | Muse Spark 1.3 latency | GPT-5.6 Luna latency |
|---|---|---|---|---|
| Vals Index | $3.38 | $0.82 | 32m42s | 29m27s |
| Harvey's Legal Agent Benchmark | $2.76 | $0.38 | 16m39s | 10m56s |
| Legal Research Bench | $0.65 | $0.85 | 8m51s | 39m34s |
| EMB | $3.55 | $0.37 | 22m55s | 11m00s |
| Finance Agent (v2) | $0.74 | $0.28 | 5m42s | 12m52s |
| Tax Agent Bench | $0.32 | $0.47 | 8m25s | 42m46s |
| ProofBench v1.1 | $0.43 | $0.12 | 8m54s | 8m34s |
| Terminal-Bench Science | $4.41 | $0.58 | 1h08m | 2h03m |
| Code Migration | $4.62 | $1.88 | 1h08m | 58m35s |
| IOI | $2.36 | $0.58 | 29m51s | 1h10m |
| ProgramBench | $11.56 | $0.94 | 1h09m | 41m34s |
| Terminal-Bench 4.0 | $7.04 | $0.73 | 1h59m | 35m55s |
| Vibe Code Bench v1.1 | $2.10 | $0.73 | 12m54s | 25m52s |
| CyberBench v1.1 | $4.26 | $0.40 | 24m53s | 16m15s |
Results available only for Muse Spark 1.3
None.
Results available only for GPT-5.6 Luna
- LegalBench
- MortgageTax
- TaxEval v2
- MedCode
- MedScribe
- BioMysteryBench
- MysteryMechanism
- GPQA Diamond
- MMLU Pro
- MMMU Pro
- SAGE
- SkillsBench
- SWE-bench
- Vibe Code Bench 1-100
- Public Benefits Bench v1.1