DeepSeek V4 Flash 0731 vs Mistral Large 4: Benchmark Comparison
DeepSeek V4 Flash 0731 has the higher score on 2 of 9 shared benchmarks; Mistral Large 4 leads on 7.
The largest observed score gap is 46.00 pts on ProofBench v1.1 , where DeepSeek V4 Flash 0731 leads.
Reported ±1 standard-error ranges overlap on 6 of 9 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 Flash 0731 | Mistral Large 4 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 30.29% ±3.19 | 31.73% ±3.23 | 1.44 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 28.07% ±3.16 | 30.39% ±3.31 | 2.32 pts | Reported ±1 SE ranges overlap |
| MedCode | 41.41% ±2.15 | 40.69% ±2.17 | 0.73 pts | Reported ±1 SE ranges overlap |
| MedScribe | 80.36% ±1.97 | 80.43% ±2.00 | 0.07 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 56.00% ±4.99 | 10.00% ±3.02 | 46.00 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 64.44% ±1.28 | 67.04% ±2.67 | 2.59 pts | Reported ±1 SE ranges overlap |
| IOI | 32.72% ±2.35 | 45.28% ±3.78 | 12.56 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 18.69% ±3.54 | 27.78% ±1.82 | 9.09 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 74.74% ±3.44 | 78.40% ±3.55 | 3.67 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | DeepSeek V4 Flash 0731 average | Mistral Large 4 average |
|---|---|---|
| Legal | 30.29% | 31.73% |
| Finance | 28.07% | 30.39% |
| Healthcare | 60.89% | 60.56% |
| Math | 56.00% | 10.00% |
| Science | 64.44% | 67.04% |
| Coding | 42.05% | 50.49% |
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 Flash 0731 cost | Mistral Large 4 cost | DeepSeek V4 Flash 0731 latency | Mistral Large 4 latency |
|---|---|---|---|---|
| Legal Research Bench | $0.28 | $5.33 | 21m07s | 1h33m |
| Tax Agent Bench | $0.16 | $2.72 | 5m37s | 57m18s |
| MedCode | N/A | N/A | 2m07s | 68.68s |
| MedScribe | N/A | N/A | 77.55s | 2m05s |
| ProofBench v1.1 | $0.04 | $0.64 | 15m34s | 20m59s |
| BioMysteryBench | $0.04 | $2.07 | 27m03s | 25m32s |
| IOI | $0.37 | $5.39 | 27m43s | 1h24m |
| Terminal-Bench 4.0 | $0.77 | $10.25 | 1h46m | 1h40m |
| Vibe Code Bench v1.1 | $0.91 | $39.46 | 37m22s | 2h53m |
Results available only for DeepSeek V4 Flash 0731
- LegalBench
- TaxEval v2
- Terminal-Bench-Science
- GPQA Diamond
- MMLU Pro
- LiveCodeBench
- ProgramBench
- SkillsBench
- SWE-bench
- CyberBench v1.1
Results available only for Mistral Large 4
- Vals Finance Index
- Vals Legal Index
- HLAB-Vals-1.1
- EMB v1.1
- Finance Agent v2.1
- MysteryMechanism
- SWE-bench Multimodal
- Vibe Code Bench 1-100