Claude Haiku 5.5 vs GLM 5.3: Benchmark Comparison
Claude Haiku 5.5 has the higher score on 10 of 22 shared benchmarks; GLM 5.3 leads on 12.
The largest observed score gap is 40.00 pts on ProofBench v1.1 , where Claude Haiku 5.5 leads.
Reported ±1 standard-error ranges overlap on 12 of 22 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 Haiku 5.5 | GLM 5.3 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Finance Index | 57.45% ±1.80 | 60.54% ±1.52 | 3.10 pts | Reported ±1 SE ranges overlap |
| Vals Index | 58.71% ±1.28 | 59.73% ±1.11 | 1.02 pts | Reported ±1 SE ranges overlap |
| Vals Legal Index | 70.27% ±1.83 | 74.58% ±1.84 | 4.31 pts | Reported ±1 SE ranges do not overlap |
| HLAB-Vals-1.1 | 4.17% ±1.52 | 14.58% ±2.86 | 10.42 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 43.27% ±3.44 | 49.04% ±3.48 | 5.77 pts | Reported ±1 SE ranges overlap |
| EMB v1.1 | 65.48% ±3.69 | 69.11% ±2.65 | 3.63 pts | Reported ±1 SE ranges overlap |
| Finance Agent v2.1 | 65.37% ±2.06 | 65.00% ±2.01 | 0.37 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 31.92% ±3.35 | 39.70% ±3.52 | 7.78 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 51.54% ±2.17 | 42.86% ±2.11 | 8.68 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 89.54% ±1.92 | 88.81% ±2.00 | 0.73 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 89.00% ±3.15 | 49.00% ±5.02 | 40.00 pts | Reported ±1 SE ranges do not overlap |
| MysteryMechanism | 35.13% ±3.21 | 22.97% ±2.83 | 12.16 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench-Science | 10.00% ±3.61 | 5.71% ±2.79 | 4.29 pts | Reported ±1 SE ranges overlap |
| Code Migration v1.1 | 52.36% ±3.49 | 55.07% ±2.12 | 2.71 pts | Reported ±1 SE ranges overlap |
| IOI | 47.28% ±10.72 | 68.44% ±7.48 | 21.17 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 2.50% ±1.11 | 1.50% ±0.86 | 1.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 36.36% ±2.31 | 38.89% ±1.82 | 2.53 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench 1-100 | 20.41% ±3.79 | 19.99% ±3.81 | 0.41 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 90.44% ±1.73 | 78.13% ±3.94 | 12.32 pts | Reported ±1 SE ranges do not overlap |
| CyberBench v1.1 | 75.48% ±5.21 | 72.08% ±5.41 | 3.39 pts | Reported ±1 SE ranges overlap |
| SRE Bench | 5.72% ±1.44 | 9.16% ±1.79 | 3.44 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 63.19% ±1.25 | 68.54% ±1.21 | 5.34 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Haiku 5.5 average | GLM 5.3 average |
|---|---|---|
| Legal | 23.72% | 31.81% |
| Finance | 54.26% | 57.94% |
| Healthcare | 70.54% | 65.84% |
| Math | 89.00% | 49.00% |
| Science | 22.57% | 14.34% |
| Coding | 41.56% | 43.67% |
| Cyber | 40.60% | 40.62% |
| Social Mobility | 63.19% | 68.54% |
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 Haiku 5.5 cost | GLM 5.3 cost | Claude Haiku 5.5 latency | GLM 5.3 latency |
|---|---|---|---|---|
| Vals Finance Index | $1.39 | $2.21 | 33m31s | 35m44s |
| Vals Index | $3.17 | $8.52 | 49m48s | 1h04m |
| Vals Legal Index | $2.35 | $2.68 | 58m22s | 44m51s |
| HLAB-Vals-1.1 | $4.26 | $3.99 | 43m41s | 31m51s |
| Legal Research Bench | $1.63 | $2.24 | 1h22m | 51m50s |
| EMB v1.1 | $1.36 | $3.79 | 19m56s | 36m12s |
| Finance Agent v2.1 | $1.64 | $1.03 | 31m05s | 20m06s |
| Tax Agent Bench | $1.16 | $1.80 | 49m34s | 50m53s |
| MedCode | N/A | N/A | 3m46s | 2m49s |
| MedScribe | N/A | N/A | 5m23s | 2m02s |
| ProofBench v1.1 | $0.17 | $2.08 | 6m58s | 42m23s |
| MysteryMechanism | $0.54 | $1.53 | 13m25s | 30m42s |
| Terminal-Bench-Science | $4.52 | $15.63 | 1h34m | 2h45m |
| Code Migration v1.1 | $6.20 | $34.91 | 1h26m | 2h58m |
| IOI | $0.83 | $7.67 | 12m34s | 1h46m |
| ProgramBench | $6.83 | $21.96 | 1h13m | 4h16m |
| Terminal-Bench 4.0 | $3.12 | $9.37 | 51m56s | 1h23m |
| Vibe Code Bench 1-100 | $8.30 | $8.24 | 4h06m | 1h26m |
| Vibe Code Bench v1.1 | $6.07 | $12.45 | 34m36s | 1h04m |
| CyberBench v1.1 | $1.25 | $2.69 | 32m39s | 30m18s |
| SRE Bench | $3.14 | $27.47 | 57m56s | 2h48m |
| Public Benefits Bench v1.1 | $0.53 | $0.95 | 36m32s | 35m51s |
Results available only for Claude Haiku 5.5
- Vals Science Index
- ClaimBench
- BioMysteryBench
- SAGE
- SWE-bench Multimodal
Results available only for GLM 5.3
- Vals RSI Index
- LegalBench
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- LiveCodeBench
- SkillsBench
- SWE-bench