DeepSeek V4 Flash 0731 vs GLM 5.3: Benchmark Comparison
DeepSeek V4 Flash 0731 has the higher score on 5 of 23 shared benchmarks; GLM 5.3 leads on 16.
The largest observed score gap is 35.72 pts on IOI , where GLM 5.3 leads.
Reported ±1 standard-error ranges overlap on 12 of 23 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 | GLM 5.3 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 47.96% ±1.24 | 53.51% ±1.30 | 5.55 pts | Reported ±1 SE ranges do not overlap |
| Harvey's Legal Agent Benchmark | 8.33% ±2.15 | 8.33% ±2.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 30.29% ±3.19 | 49.04% ±3.48 | 18.75 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 77.71% ±0.50 | 84.84% ±0.40 | 7.12 pts | Reported ±1 SE ranges do not overlap |
| EMB | 56.98% ±3.39 | 56.34% ±3.32 | 0.63 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 49.52% ±0.47 | 55.84% ±2.07 | 6.32 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 62.56% ±3.14 | 73.09% ±2.97 | 10.53 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 70.69% ±0.90 | 72.36% ±0.88 | 1.68 pts | Reported ±1 SE ranges overlap |
| MedCode | 41.41% ±2.15 | 42.86% ±2.11 | 1.45 pts | Reported ±1 SE ranges overlap |
| MedScribe | 80.36% ±1.97 | 88.81% ±2.00 | 8.45 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 56.00% ±4.99 | 49.00% ±5.02 | 7.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench Science | 4.29% ±2.44 | 5.71% ±2.79 | 1.43 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 89.90% ±1.69 | 88.13% ±1.85 | 1.77 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 86.21% ±0.34 | 86.77% ±0.34 | 0.56 pts | Reported ±1 SE ranges overlap |
| Code Migration | 38.63% ±4.38 | 44.22% ±4.29 | 5.59 pts | Reported ±1 SE ranges overlap |
| IOI | 32.72% ±2.35 | 68.44% ±7.48 | 35.72 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 87.26% ±0.97 | 80.53% ±1.06 | 6.73 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 1.50% ±0.86 | 1.50 pts | Reported ±1 SE ranges do not overlap |
| SkillsBench | 50.67% ±4.43 | 47.51% ±4.52 | 3.16 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 88.80% ±1.41 | 95.40% ±0.94 | 6.60 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 18.69% ±3.54 | 38.89% ±1.82 | 20.20 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 74.74% ±3.44 | 78.13% ±3.94 | 3.39 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 72.08% ±5.41 | 72.08% ±5.41 | 0.00 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | DeepSeek V4 Flash 0731 average | GLM 5.3 average |
|---|---|---|
| Legal | 38.78% | 47.40% |
| Finance | 59.94% | 64.41% |
| Healthcare | 60.89% | 65.84% |
| Math | 56.00% | 49.00% |
| Science | 4.29% | 5.71% |
| Academic | 88.05% | 87.45% |
| Coding | 48.94% | 56.83% |
| Cyber | 72.08% | 72.08% |
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 | GLM 5.3 cost | DeepSeek V4 Flash 0731 latency | GLM 5.3 latency |
|---|---|---|---|---|
| Vals Index | $0.85 | $7.25 | 43m06s | 1h13m |
| Harvey's Legal Agent Benchmark | $0.06 | $4.30 | 16m51s | 42m38s |
| Legal Research Bench | $0.28 | $2.24 | 21m07s | 51m50s |
| LegalBench | N/A | N/A | 3.95s | 22.48s |
| EMB | $0.38 | $3.79 | 26m14s | 36m12s |
| Finance Agent (v2) | $0.28 | $1.07 | 8m10s | 15m50s |
| Tax Agent Bench | $0.16 | $1.80 | 5m37s | 50m53s |
| TaxEval v2 | N/A | N/A | 72.20s | 98.32s |
| MedCode | N/A | N/A | 2m07s | 2m49s |
| MedScribe | N/A | N/A | 77.55s | 2m02s |
| ProofBench v1.1 | $0.04 | $2.08 | 15m34s | 42m23s |
| Terminal-Bench Science | $1.10 | $15.63 | 1h34m | 2h45m |
| GPQA Diamond | N/A | N/A | 79.96s | 3m31s |
| MMLU Pro | N/A | N/A | 17.19s | 61.78s |
| Code Migration | $3.61 | $24.91 | 2h08m | 3h47m |
| IOI | $0.37 | $7.67 | 27m43s | 1h46m |
| LiveCodeBench | N/A | N/A | 2m07s | 4m07s |
| ProgramBench | $1.29 | $21.96 | 5h39m | 4h16m |
| SkillsBench | $0.13 | $0.71 | 6m52s | 13m07s |
| SWE-bench | $0.01 | $0.34 | 3m22s | 14m01s |
| Terminal-Bench 4.0 | $0.77 | $9.37 | 1h46m | 1h23m |
| Vibe Code Bench v1.1 | $0.91 | $12.45 | 37m22s | 1h04m |
| CyberBench v1.1 | $0.19 | $2.69 | 15m28s | 30m18s |
Results available only for DeepSeek V4 Flash 0731
- BioMysteryBench
Results available only for GLM 5.3
- Vals RSI Index
- MysteryMechanism
- Vibe Code Bench 1-100
- Public Benefits Bench v1.1