Gemini 3.6 Flash vs Hy4 Preview: Benchmark Comparison
Gemini 3.6 Flash has the higher score on 5 of 16 shared benchmarks; Hy4 Preview leads on 10.
The largest observed score gap is 24.28 pts on IOI , where Hy4 Preview leads.
Reported ±1 standard-error ranges overlap on 2 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 | Gemini 3.6 Flash | Hy4 Preview | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 3.33% ±1.17 | 9.17% ±2.00 | 5.83 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 25.00% ±3.01 | 45.19% ±3.46 | 20.19 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 86.70% ±0.41 | 83.76% ±0.41 | 2.95 pts | Reported ±1 SE ranges do not overlap |
| EMB | 65.41% ±2.65 | 58.08% ±3.14 | 7.34 pts | Reported ±1 SE ranges do not overlap |
| Finance Agent (v2) | 56.30% ±0.18 | 55.06% ±0.31 | 1.23 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 49.90% ±3.30 | 63.71% ±3.26 | 13.82 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 53.15% ±2.16 | 43.25% ±2.13 | 9.91 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 79.66% ±1.86 | 83.60% ±2.06 | 3.94 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 58.52% ±1.61 | 69.26% ±2.59 | 10.74 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench Science | 4.29% ±2.44 | 1.43% ±1.43 | 2.86 pts | Reported ±1 SE ranges overlap |
| Code Migration | 30.93% ±4.07 | 47.43% ±4.27 | 16.51 pts | Reported ±1 SE ranges do not overlap |
| IOI | 35.06% ±5.70 | 59.33% ±4.60 | 24.28 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 64.00% ±4.29 | 77.48% ±4.04 | 13.47 pts | Reported ±1 SE ranges do not overlap |
| CyberBench v1.1 | 45.36% ±3.75 | 65.36% ±5.55 | 20.00 pts | Reported ±1 SE ranges do not overlap |
| Public Benefits Bench v1.1 | 56.56% ±1.29 | 68.61% ±1.21 | 12.04 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Gemini 3.6 Flash average | Hy4 Preview average |
|---|---|---|
| Legal | 38.35% | 46.04% |
| Finance | 57.20% | 58.95% |
| Healthcare | 66.41% | 63.42% |
| Science | 31.40% | 35.34% |
| Coding | 32.50% | 46.06% |
| Cyber | 45.36% | 65.36% |
| Social Mobility | 56.56% | 68.61% |
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 | Gemini 3.6 Flash cost | Hy4 Preview cost | Gemini 3.6 Flash latency | Hy4 Preview latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $1.90 | $0.98 | 16m24s | 29m25s |
| Legal Research Bench | $0.47 | $0.85 | 2m48s | 1h04m |
| LegalBench | N/A | N/A | 2.96s | 90.56s |
| EMB | $3.89 | $0.79 | 12m06s | 30m39s |
| Finance Agent (v2) | $1.40 | $0.59 | 3m45s | 23m45s |
| Tax Agent Bench | $0.36 | $0.59 | 105.32s | 41m04s |
| MedCode | N/A | N/A | 16.68s | 6m39s |
| MedScribe | N/A | N/A | 33.55s | 5m50s |
| BioMysteryBench | $1.25 | $0.36 | 12m42s | 23m30s |
| Terminal-Bench Science | $4.39 | $2.14 | 1h52m | 3h48m |
| Code Migration | $8.93 | $3.41 | 1h30m | 2h13m |
| IOI | $8.67 | $1.35 | 48m23s | 1h01m |
| ProgramBench | $5.92 | $12.81 | 47m33s | 2h10m |
| Vibe Code Bench v1.1 | $3.04 | $2.18 | 25m23s | 41m33s |
| CyberBench v1.1 | $1.31 | $0.49 | 11m00s | 36m43s |
| Public Benefits Bench v1.1 | $0.39 | $0.25 | 2m02s | 1h03m |
Results available only for Gemini 3.6 Flash
- MortgageTax
- TaxEval v2
- GPQA Diamond
- MMLU Pro
- MMMU Pro
- SAGE
- LiveCodeBench
- SWE-bench
Results available only for Hy4 Preview
- Vals Index
- ProofBench v1.1
- Terminal-Bench 4.0
- SRE Bench