DeepSeek V4.1 Flash vs GPT-5.6 Luna: Benchmark Comparison
DeepSeek V4.1 Flash has the higher score on 13 of 20 shared benchmarks; GPT-5.6 Luna leads on 6.
The largest observed score gap is 21.50 pts on IOI , where GPT-5.6 Luna leads.
Reported ±1 standard-error ranges overlap on 12 of 20 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 | DeepSeek V4.1 Flash | GPT-5.6 Luna | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Science Index | 29.65% ±0.99 | 25.30% ±0.82 | 4.35 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 41.35% ±3.42 | 36.54% ±3.35 | 4.81 pts | Reported ±1 SE ranges overlap |
| LegalBench | 83.28% ±0.46 | 84.03% ±0.42 | 0.76 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 26.63% ±3.14 | 24.82% ±3.11 | 1.81 pts | Reported ±1 SE ranges overlap |
| MedCode | 41.17% ±2.04 | 42.39% ±2.27 | 1.22 pts | Reported ±1 SE ranges overlap |
| MedScribe | 85.50% ±1.92 | 84.39% ±2.58 | 1.11 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 54.00% ±5.01 | 60.00% ±4.92 | 6.00 pts | Reported ±1 SE ranges overlap |
| BioMysteryBench | 67.78% ±1.11 | 61.48% ±0.74 | 6.30 pts | Reported ±1 SE ranges do not overlap |
| MysteryMechanism | 21.17% ±2.75 | 14.41% ±2.36 | 6.76 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench-Science | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| SAGE | 47.88% ±3.44 | 44.22% ±3.33 | 3.66 pts | Reported ±1 SE ranges overlap |
| IOI | 40.28% ±2.56 | 61.78% ±11.61 | 21.50 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 |
| SkillsBench | 69.80% ±3.88 | 60.45% ±4.75 | 9.36 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 19.70% ±1.75 | 11.62% ±1.01 | 8.08 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench 1-100 | 16.38% ±3.19 | 22.59% ±4.00 | 6.22 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 84.74% ±2.89 | 77.06% ±3.07 | 7.68 pts | Reported ±1 SE ranges do not overlap |
| CyberBench v1.1 | 73.69% ±5.48 | 73.63% ±5.56 | 0.06 pts | Reported ±1 SE ranges overlap |
| SRE Bench | 0.76% ±0.54 | 1.53% ±0.76 | 0.76 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 64.28% ±1.25 | 61.16% ±1.27 | 3.11 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | DeepSeek V4.1 Flash average | GPT-5.6 Luna average |
|---|---|---|
| Legal | 62.31% | 60.29% |
| Finance | 26.63% | 24.82% |
| Healthcare | 63.34% | 63.39% |
| Math | 54.00% | 60.00% |
| Science | 29.65% | 25.30% |
| Education | 47.88% | 44.22% |
| Coding | 38.57% | 38.91% |
| Cyber | 37.23% | 37.58% |
| Social Mobility | 64.28% | 61.16% |
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 | DeepSeek V4.1 Flash cost | GPT-5.6 Luna cost | DeepSeek V4.1 Flash latency | GPT-5.6 Luna latency |
|---|---|---|---|---|
| Vals Science Index | $0.29 | $0.26 | 1h07m | 50m00s |
| Legal Research Bench | $0.25 | $0.85 | 11m42s | 39m34s |
| LegalBench | N/A | N/A | 6.53s | 5.89s |
| Tax Agent Bench | $0.13 | $0.47 | 5m49s | 42m46s |
| MedCode | N/A | N/A | 35.13s | 81.28s |
| MedScribe | N/A | N/A | 53.83s | 116.89s |
| ProofBench v1.1 | $0.13 | $0.12 | 9m07s | 8m34s |
| BioMysteryBench | $0.23 | $0.09 | 14m50s | 18m31s |
| MysteryMechanism | $0.16 | $0.11 | 8m04s | 8m13s |
| Terminal-Bench-Science | $0.47 | $0.58 | 2h57m | 2h03m |
| SAGE | N/A | N/A | 35.78s | 51.88s |
| IOI | $0.24 | $0.58 | 18m30s | 1h10m |
| ProgramBench | $0.90 | $0.94 | 1h02m | 41m34s |
| SkillsBench | $0.09 | $0.21 | 3m43s | 5m20s |
| Terminal-Bench 4.0 | $0.50 | $0.73 | 48m22s | 35m55s |
| Vibe Code Bench 1-100 | $0.69 | $4.91 | 28m21s | 2h03m |
| Vibe Code Bench v1.1 | $0.41 | $0.73 | 15m28s | 25m52s |
| CyberBench v1.1 | $0.07 | $0.40 | 13m50s | 16m15s |
| SRE Bench | $0.55 | $3.44 | 1h01m | 1h42m |
| Public Benefits Bench v1.1 | $0.07 | $0.27 | 9m59s | 38m40s |
Results available only for DeepSeek V4.1 Flash
- Vals Finance Index
- Vals Inference Provider Index
- EMB v1.1
- Finance Agent v2.1
- Code Migration v1.1
- SWE-bench Multimodal
Results available only for GPT-5.6 Luna
- MortgageTax
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- MMMU Pro
- SWE-bench