DeepSeek V4 Pro 0813 vs GLM 5.2: Benchmark Comparison
DeepSeek V4 Pro 0813 has the higher score on 11 of 16 shared benchmarks; GLM 5.2 leads on 5.
The largest observed score gap is 18.34 pts on Vibe Code Bench v1.1 , where DeepSeek V4 Pro 0813 leads.
Reported ±1 standard-error ranges overlap on 9 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 Pro 0813 | GLM 5.2 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 7.50% ±2.15 | 7.08% ±2.00 | 0.42 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 40.87% ±3.42 | 31.25% ±3.22 | 9.62 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 82.36% ±0.44 | 84.07% ±0.45 | 1.71 pts | Reported ±1 SE ranges do not overlap |
| EMB | 52.80% ±3.06 | 61.53% ±2.88 | 8.73 pts | Reported ±1 SE ranges do not overlap |
| Finance Agent (v2) | 50.39% ±0.25 | 49.70% ±0.88 | 0.69 pts | Reported ±1 SE ranges overlap |
| TaxEval v2 | 73.06% ±0.87 | 73.34% ±0.86 | 0.29 pts | Reported ±1 SE ranges overlap |
| MedCode | 42.47% ±2.16 | 40.77% ±2.17 | 1.70 pts | Reported ±1 SE ranges overlap |
| MedScribe | 80.17% ±2.00 | 83.53% ±2.00 | 3.36 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 92.42% ±2.02 | 85.61% ±1.77 | 6.82 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 86.97% ±0.34 | 86.71% ±0.33 | 0.26 pts | Reported ±1 SE ranges overlap |
| Code Migration | 41.54% ±4.30 | 37.87% ±4.14 | 3.67 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 87.53% ±0.96 | 69.50% ±1.18 | 18.03 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 0.50% ±0.50 | 0.50 pts | Reported ±1 SE ranges overlap |
| SkillsBench | 53.83% ±4.49 | 45.08% ±4.39 | 8.75 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 96.40% ±0.83 | 82.80% ±1.69 | 13.60 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 82.30% ±3.15 | 63.96% ±4.79 | 18.34 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | DeepSeek V4 Pro 0813 average | GLM 5.2 average |
|---|---|---|
| Legal | 43.58% | 40.80% |
| Finance | 58.75% | 61.52% |
| Healthcare | 61.32% | 62.15% |
| Academic | 89.70% | 86.16% |
| Coding | 60.27% | 49.95% |
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 Pro 0813 cost | GLM 5.2 cost | DeepSeek V4 Pro 0813 latency | GLM 5.2 latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $0.17 | $2.06 | 21m53s | 20m51s |
| Legal Research Bench | $1.09 | $0.89 | 42m07s | 17m03s |
| LegalBench | N/A | N/A | 20.05s | 5.77s |
| EMB | $1.08 | $4.47 | 37m27s | 58m52s |
| Finance Agent (v2) | $0.88 | $0.71 | 16m15s | 7m53s |
| TaxEval v2 | N/A | N/A | 3m46s | 64.94s |
| MedCode | N/A | N/A | 3m01s | 95.27s |
| MedScribe | N/A | N/A | 2m36s | 2m18s |
| GPQA Diamond | N/A | N/A | 4m39s | 3m45s |
| MMLU Pro | N/A | N/A | 59.62s | 51.64s |
| Code Migration | $18.58 | $12.64 | 3h32m | 1h54m |
| LiveCodeBench | N/A | N/A | 4m33s | 8m52s |
| ProgramBench | $0.43 | $12.58 | 1h23m | 2h06m |
| SkillsBench | $0.37 | $0.81 | 11m15s | 19m52s |
| SWE-bench | $0.10 | $0.71 | 4m00s | 11m10s |
| Vibe Code Bench v1.1 | $0.36 | $8.49 | 1h06m | 1h02m |
Results available only for DeepSeek V4 Pro 0813
- Vals Index
- Tax Agent Bench
- ProofBench v1.1
- Terminal-Bench Science
- IOI
- Terminal-Bench 4.0
- Vibe Code Bench 1-100
- CyberBench v1.1
Results available only for GLM 5.2
- SRE Bench