Model comparison
Claude Sonnet 4.6 vs GLM 5.2: Benchmark Comparison
Claude Sonnet 4.6 has the higher score on 7 of 14 shared benchmarks; GLM 5.2 leads on 5.
The largest observed score gap is 12.59 pts on LiveCodeBench , where Claude Sonnet 4.6 leads.
Reported ±1 standard-error ranges overlap on 7 of 14 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 | Claude Sonnet 4.6 | GLM 5.2 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 5.00% ±1.65 | 7.08% ±2.00 | 2.08 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 38.46% ±3.38 | 31.25% ±3.22 | 7.21 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 82.12% ±0.48 | 84.07% ±0.45 | 1.95 pts | Reported ±1 SE ranges do not overlap |
| EMB | 60.15% ±3.28 | 61.53% ±2.88 | 1.38 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 51.03% ±0.33 | 49.70% ±0.88 | 1.34 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 77.11% ±0.82 | 73.34% ±0.86 | 3.76 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 85.61% ±2.41 | 85.61% ±1.77 | 0.00 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 87.34% ±0.43 | 86.71% ±0.33 | 0.63 pts | Reported ±1 SE ranges overlap |
| Code Migration | 39.89% ±4.14 | 37.87% ±4.14 | 2.02 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 82.09% ±1.06 | 69.50% ±1.18 | 12.59 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.50% ±0.50 | 0.50% ±0.50 | 0.00 pts | Reported ±1 SE ranges overlap |
| SkillsBench | 49.05% ±4.64 | 45.08% ±4.39 | 3.98 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 77.40% ±1.87 | 82.80% ±1.69 | 5.40 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 51.48% ±4.64 | 63.96% ±4.79 | 12.48 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Sonnet 4.6 average | GLM 5.2 average |
|---|---|---|
| Legal | 41.86% | 40.80% |
| Finance | 62.76% | 61.52% |
| Academic | 86.47% | 86.16% |
| Coding | 50.07% | 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 | Claude Sonnet 4.6 cost | GLM 5.2 cost | Claude Sonnet 4.6 latency | GLM 5.2 latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $3.04 | $2.06 | 18m43s | 20m51s |
| Legal Research Bench | $2.30 | $0.89 | 16m30s | 17m03s |
| LegalBench | N/A | N/A | 21.63s | 5.77s |
| EMB | $7.23 | $4.47 | 54m16s | 58m52s |
| Finance Agent (v2) | $2.41 | $0.71 | 11m36s | 7m53s |
| TaxEval v2 | N/A | N/A | 2m08s | 64.94s |
| GPQA Diamond | N/A | N/A | 6m18s | 3m45s |
| MMLU Pro | N/A | N/A | 55.41s | 51.64s |
| Code Migration | $38.51 | $12.64 | 1h40m | 1h54m |
| LiveCodeBench | N/A | N/A | 4m55s | 8m52s |
| ProgramBench | N/A | N/A | 1h12m | 2h06m |
| SkillsBench | $1.38 | $0.81 | 14m20s | 19m52s |
| SWE-bench | $1.30 | $0.71 | 8m32s | 11m10s |
| Vibe Code Bench v1.1 | $5.91 | $8.49 | 26m12s | 1h02m |
Results available only for Claude Sonnet 4.6
- MortgageTax
- Tax Agent Bench
- MMMU Pro
- SAGE
- Public Benefits Bench v1.1
Results available only for GLM 5.2
- MedCode
- MedScribe
- SRE Bench