DeepSeek V4 vs Inkling Small: Benchmark Comparison
DeepSeek V4 has the higher score on 10 of 16 shared benchmarks; Inkling Small leads on 6.
The largest observed score gap is 30.88 pts on Vibe Code Bench v1.1 , where DeepSeek V4 leads.
Reported ±1 standard-error ranges overlap on 4 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 Small | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 23.08% ±2.93 | 25.48% ±3.03 | 2.40 pts | Reported ±1 SE ranges overlap |
| LegalBench | 80.32% ±0.47 | 82.99% ±0.46 | 2.66 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 25.17% ±3.06 | 15.39% ±2.37 | 9.78 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 72.08% ±0.88 | 75.51% ±0.84 | 3.43 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 40.45% ±2.12 | 37.89% ±2.21 | 2.56 pts | Reported ±1 SE ranges overlap |
| MedScribe | 75.14% ±2.00 | 84.11% ±1.87 | 8.97 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 16.00% ±3.69 | 6.00% ±2.39 | 10.00 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 89.39% ±1.65 | 83.59% ±2.04 | 5.81 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 87.25% ±0.34 | 85.57% ±0.35 | 1.68 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 87.48% ±0.95 | 85.93% ±0.99 | 1.55 pts | Reported ±1 SE ranges overlap |
| ProgramBench | 0.00% ±0.00 | 0.50% ±0.50 | 0.50 pts | Reported ±1 SE ranges overlap |
| SkillsBench | 51.27% ±4.60 | 33.58% ±3.66 | 17.69 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench | 77.40% ±1.87 | 82.20% ±1.71 | 4.80 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 11.11% ±1.34 | 0.00% ±0.00 | 11.11 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 49.93% ±4.77 | 19.05% ±3.40 | 30.88 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 62.92% ±1.26 | 59.41% ±1.28 | 3.52 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | DeepSeek V4 average | Inkling Small average |
|---|---|---|
| Legal | 51.70% | 54.23% |
| Finance | 48.62% | 45.45% |
| Healthcare | 57.80% | 61.00% |
| Math | 16.00% | 6.00% |
| Academic | 88.32% | 84.58% |
| Coding | 46.20% | 36.88% |
| Social Mobility | 62.92% | 59.41% |
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 Small cost | DeepSeek V4 latency | Inkling Small latency |
|---|---|---|---|---|
| Legal Research Bench | $0.68 | $0.14 | 17m05s | 12m55s |
| LegalBench | N/A | N/A | 3m04s | 17.66s |
| Tax Agent Bench | $0.72 | $0.09 | 20m45s | 6m25s |
| TaxEval v2 | N/A | N/A | 3m10s | 4m02s |
| MedCode | N/A | N/A | 6m23s | 3m11s |
| MedScribe | N/A | N/A | 5m46s | 3m50s |
| ProofBench v1.1 | $0.03 | $0.08 | 7m03s | 8m27s |
| GPQA Diamond | N/A | N/A | 15m00s | 4m20s |
| MMLU Pro | N/A | N/A | 9m38s | 71.69s |
| LiveCodeBench | N/A | N/A | 11m56s | 5m02s |
| ProgramBench | N/A | N/A | 1h03m | 28m31s |
| SkillsBench | $0.40 | $0.15 | 8m16s | 11m41s |
| SWE-bench | $0.44 | $0.13 | 10m35s | 4m05s |
| Terminal-Bench 4.0 | $3.45 | $0.32 | 1h26m | 11m45s |
| Vibe Code Bench v1.1 | $2.21 | $0.43 | 56m59s | 49m09s |
| Public Benefits Bench v1.1 | $0.46 | $0.10 | 15m22s | 8m41s |
Results available only for DeepSeek V4
None.
Results available only for Inkling Small
- Vals Finance Index
- Vals Index
- HLAB-Vals-1.1
- EMB v1.1
- Finance Agent v2.1
- MortgageTax
- SAGE
- Code Migration v1.1
- IOI
- CyberBench v1.1