Claude Haiku 5.5 vs Grok 4.6: Benchmark Comparison
Claude Haiku 5.5 has the higher score on 11 of 17 shared benchmarks; Grok 4.6 leads on 6.
The largest observed score gap is 38.00 pts on ProofBench v1.1 , where Claude Haiku 5.5 leads.
Reported ±1 standard-error ranges overlap on 10 of 17 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 | Grok 4.6 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Science Index | 38.50% ±1.67 | 38.09% ±1.64 | 0.41 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 43.27% ±3.44 | 48.08% ±3.47 | 4.81 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 31.92% ±3.35 | 33.14% ±3.40 | 1.22 pts | Reported ±1 SE ranges overlap |
| MedCode | 51.54% ±2.17 | 44.71% ±2.26 | 6.83 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 89.54% ±1.92 | 86.53% ±1.96 | 3.01 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 89.00% ±3.15 | 51.00% ±5.02 | 38.00 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 70.37% ±1.33 | 72.22% ±0.00 | 1.85 pts | Reported ±1 SE ranges do not overlap |
| MysteryMechanism | 35.13% ±3.21 | 30.63% ±3.10 | 4.50 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench-Science | 10.00% ±3.61 | 11.43% ±3.83 | 1.43 pts | Reported ±1 SE ranges overlap |
| SAGE | 45.78% ±3.42 | 28.90% ±3.08 | 16.88 pts | Reported ±1 SE ranges do not overlap |
| IOI | 47.28% ±10.72 | 47.61% ±2.09 | 0.33 pts | Reported ±1 SE ranges overlap |
| ProgramBench | 2.50% ±1.11 | 1.00% ±0.70 | 1.50 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 36.36% ±2.31 | 17.17% ±1.34 | 19.19 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench 1-100 | 20.41% ±3.79 | 14.75% ±3.17 | 5.66 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 90.44% ±1.73 | 76.24% ±3.82 | 14.20 pts | Reported ±1 SE ranges do not overlap |
| CyberBench v1.1 | 75.48% ±5.21 | 67.68% ±5.87 | 7.80 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 63.19% ±1.25 | 66.85% ±1.23 | 3.65 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Claude Haiku 5.5 average | Grok 4.6 average |
|---|---|---|
| Legal | 43.27% | 48.08% |
| Finance | 31.92% | 33.14% |
| Healthcare | 70.54% | 65.62% |
| Math | 89.00% | 51.00% |
| Science | 38.50% | 38.09% |
| Education | 45.78% | 28.90% |
| Coding | 39.40% | 31.35% |
| Cyber | 75.48% | 67.68% |
| Social Mobility | 63.19% | 66.85% |
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 | Grok 4.6 cost | Claude Haiku 5.5 latency | Grok 4.6 latency |
|---|---|---|---|---|
| Vals Science Index | $1.86 | $2.45 | 39m58s | 40m10s |
| Legal Research Bench | $1.63 | $1.53 | 1h22m | 25m04s |
| Tax Agent Bench | $1.16 | $0.98 | 49m34s | 10m58s |
| MedCode | N/A | N/A | 3m46s | 2m07s |
| MedScribe | N/A | N/A | 5m23s | 70.94s |
| ProofBench v1.1 | $0.17 | $0.76 | 6m58s | 14m22s |
| BioMysteryBench | $0.51 | $1.23 | 12m32s | 14m18s |
| MysteryMechanism | $0.54 | $1.31 | 13m25s | 27m55s |
| Terminal-Bench-Science | $4.52 | $4.82 | 1h34m | 1h18m |
| SAGE | N/A | N/A | 2m23s | 3m22s |
| IOI | $0.83 | $8.18 | 12m34s | 3h57m |
| ProgramBench | $6.83 | $16.76 | 1h13m | 4h27m |
| Terminal-Bench 4.0 | $3.12 | $5.06 | 51m56s | 29m25s |
| Vibe Code Bench 1-100 | $8.30 | $26.27 | 4h06m | 2h11m |
| Vibe Code Bench v1.1 | $6.07 | $4.88 | 34m36s | 25m30s |
| CyberBench v1.1 | $1.25 | $4.07 | 32m39s | 36m52s |
| Public Benefits Bench v1.1 | $0.53 | $0.92 | 36m32s | 25m53s |
Results available only for Claude Haiku 5.5
- Vals Finance Index
- Vals Index
- Vals Legal Index
- HLAB-Vals-1.1
- ClaimBench
- EMB v1.1
- Finance Agent v2.1
- Code Migration v1.1
- SWE-bench Multimodal
- SRE Bench
Results available only for Grok 4.6
- Vals RSI Index
- LegalBench
- MortgageTax
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- LiveCodeBench
- SkillsBench
- SWE-bench
- Time Horizon Index: KSP