DeepSeek V4 vs Inkling: Benchmark Comparison
DeepSeek V4 has the higher score on 9 of 16 shared benchmarks; Inkling leads on 6.
The largest observed score gap is 30.72 pts on Vibe Code Bench v1.1 , where DeepSeek V4 leads.
Reported ±1 standard-error ranges overlap on 5 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 | DeepSeek V4 | Inkling | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 23.08% ±2.93 | 28.36% ±3.13 | 5.29 pts | Reported ±1 SE ranges overlap |
| LegalBench | 80.32% ±0.47 | 82.95% ±0.50 | 2.63 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 25.17% ±3.06 | 15.29% ±2.42 | 9.88 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 72.08% ±0.88 | 75.31% ±0.84 | 3.23 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 40.45% ±2.12 | 41.19% ±2.23 | 0.73 pts | Reported ±1 SE ranges overlap |
| MedScribe | 75.14% ±2.00 | 85.41% ±1.84 | 10.26 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 16.00% ±3.69 | 0.00% ±0.00 | 16.00 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 89.39% ±1.65 | 87.12% ±1.78 | 2.27 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 87.25% ±0.34 | 86.30% ±0.34 | 0.95 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 87.48% ±0.95 | 85.52% ±1.00 | 1.97 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| SkillsBench | 51.27% ±4.60 | 26.09% ±3.86 | 25.18 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench | 77.40% ±1.87 | 77.60% ±1.87 | 0.20 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 11.11% ±1.34 | 0.51% ±0.51 | 10.61 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 49.93% ±4.77 | 19.21% ±3.66 | 30.72 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 62.92% ±1.26 | 58.46% ±1.28 | 4.47 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | DeepSeek V4 average | Inkling average |
|---|---|---|
| Legal | 51.70% | 55.66% |
| Finance | 48.62% | 45.30% |
| Healthcare | 57.80% | 63.30% |
| Math | 16.00% | 0.00% |
| Academic | 88.32% | 86.71% |
| Coding | 46.20% | 34.82% |
| Social Mobility | 62.92% | 58.46% |
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 cost | Inkling cost | DeepSeek V4 latency | Inkling latency |
|---|---|---|---|---|
| Legal Research Bench | $0.68 | $0.49 | 17m05s | 14m29s |
| LegalBench | N/A | N/A | 3m04s | 23.14s |
| Tax Agent Bench | $0.72 | $0.35 | 20m45s | 13m31s |
| TaxEval v2 | N/A | N/A | 3m10s | 3m18s |
| MedCode | N/A | N/A | 6m23s | 2m47s |
| MedScribe | N/A | N/A | 5m46s | 4m05s |
| ProofBench v1.1 | $0.03 | $0.20 | 7m03s | 7m16s |
| GPQA Diamond | N/A | N/A | 15m00s | 5m09s |
| MMLU Pro | N/A | N/A | 9m38s | 73.26s |
| LiveCodeBench | N/A | N/A | 11m56s | 4m18s |
| ProgramBench | N/A | N/A | 1h03m | 45m34s |
| SkillsBench | $0.40 | $0.83 | 8m16s | 15m39s |
| SWE-bench | $0.44 | $0.59 | 10m35s | 6m34s |
| Terminal-Bench 4.0 | $3.45 | $2.50 | 1h26m | 22m59s |
| Vibe Code Bench v1.1 | $2.21 | $1.61 | 56m59s | 18m25s |
| Public Benefits Bench v1.1 | $0.46 | $0.37 | 15m22s | 13m31s |
Results available only for DeepSeek V4
None.
Results available only for Inkling
- Vals Finance Index
- Vals Index
- HLAB-Vals-1.1
- EMB v1.1
- Finance Agent v2.1
- MortgageTax
- MMMU Pro
- SAGE
- Code Migration v1.1
- IOI
- Vibe Code Bench 1-100
- CyberBench v1.1
- SRE Bench