Gemini 3.6 Flash vs Grok 4.5: Benchmark Comparison
Gemini 3.6 Flash has the higher score on 11 of 21 shared benchmarks; Grok 4.5 leads on 9.
The largest observed score gap is 26.59 pts on MMMU Pro , where Gemini 3.6 Flash leads.
Reported ±1 standard-error ranges overlap on 8 of 21 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 | Grok 4.5 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 3.33% ±1.17 | 12.92% ±2.53 | 9.58 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 25.00% ±3.01 | 37.98% ±3.37 | 12.98 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 86.70% ±0.41 | 85.97% ±0.41 | 0.74 pts | Reported ±1 SE ranges overlap |
| EMB | 65.41% ±2.65 | 52.91% ±3.11 | 12.50 pts | Reported ±1 SE ranges do not overlap |
| Finance Agent (v2) | 56.30% ±0.18 | 48.35% ±0.46 | 7.95 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 67.61% ±0.92 | 61.80% ±0.97 | 5.80 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 49.90% ±3.30 | 64.32% ±3.16 | 14.43 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 74.86% ±0.85 | 71.67% ±0.89 | 3.19 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 53.15% ±2.16 | 43.29% ±2.31 | 9.86 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 79.66% ±1.86 | 86.88% ±1.94 | 7.22 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 93.43% ±1.33 | 92.93% ±1.29 | 0.50 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 89.28% ±0.30 | 89.22% ±0.31 | 0.06 pts | Reported ±1 SE ranges overlap |
| MMMU Pro | 88.38% ±0.77 | 61.79% ±1.17 | 26.59 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 48.69% ±3.38 | 34.97% ±3.32 | 13.72 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 30.93% ±4.07 | 36.60% ±4.14 | 5.67 pts | Reported ±1 SE ranges overlap |
| IOI | 35.06% ±5.70 | 40.56% ±3.95 | 5.50 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 88.08% ±0.94 | 87.35% ±0.96 | 0.73 pts | Reported ±1 SE ranges overlap |
| ProgramBench | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 79.60% ±1.80 | 86.60% ±1.52 | 7.00 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 64.00% ±4.29 | 69.00% ±4.48 | 5.00 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 45.36% ±3.75 | 71.01% ±5.79 | 25.66 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Gemini 3.6 Flash average | Grok 4.5 average |
|---|---|---|
| Legal | 38.35% | 45.62% |
| Finance | 62.81% | 59.81% |
| Healthcare | 66.41% | 65.09% |
| Academic | 90.36% | 81.31% |
| Education | 48.69% | 34.97% |
| Coding | 49.61% | 53.35% |
| Cyber | 45.36% | 71.01% |
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 | Grok 4.5 cost | Gemini 3.6 Flash latency | Grok 4.5 latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $1.90 | $2.02 | 16m24s | 10m15s |
| Legal Research Bench | $0.47 | $1.09 | 2m48s | 15m54s |
| LegalBench | N/A | N/A | 2.96s | 67.88s |
| EMB | $3.89 | $1.55 | 12m06s | 18m28s |
| Finance Agent (v2) | $1.40 | $1.16 | 3m45s | 4m49s |
| MortgageTax | N/A | N/A | 9.24s | 117.46s |
| Tax Agent Bench | $0.36 | $0.47 | 105.32s | 4m44s |
| TaxEval v2 | N/A | N/A | 12.34s | 3m15s |
| MedCode | N/A | N/A | 16.68s | 59.64s |
| MedScribe | N/A | N/A | 33.55s | 11m45s |
| GPQA Diamond | N/A | N/A | 18.06s | 3m38s |
| MMLU Pro | N/A | N/A | 7.30s | 5m58s |
| MMMU Pro | N/A | N/A | 11.31s | 11m55s |
| SAGE | N/A | N/A | 21.96s | 2m50s |
| Code Migration | $8.93 | $3.37 | 1h30m | 23m29s |
| IOI | $8.67 | $5.64 | 48m23s | 1h27m |
| LiveCodeBench | N/A | N/A | 31.88s | 9m49s |
| ProgramBench | $5.92 | $4.12 | 47m33s | 18m59s |
| SWE-bench | $1.19 | $0.54 | 13m08s | 3m20s |
| Vibe Code Bench v1.1 | $3.04 | $4.10 | 25m23s | 13m50s |
| CyberBench v1.1 | $1.31 | $2.95 | 11m00s | 14m08s |
Results available only for Gemini 3.6 Flash
- BioMysteryBench
- Terminal-Bench Science
- Public Benefits Bench v1.1
Results available only for Grok 4.5
- Vals Index
- ProofBench v1.1
- SkillsBench
- Terminal-Bench 4.0
- SRE Bench
- Time Horizon Index: KSP