Gemini 3.1 Pro Preview (02/26) vs Grok 4.5: Benchmark Comparison
Gemini 3.1 Pro Preview (02/26) has the higher score on 10 of 19 shared benchmarks; Grok 4.5 leads on 8.
The largest observed score gap is 36.97 pts on Vibe Code Bench v1.1 , where Grok 4.5 leads.
Reported ±1 standard-error ranges overlap on 4 of 19 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 3.1 Pro Preview (02/26) | Grok 4.5 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 20.67% ±2.81 | 37.98% ±3.37 | 17.31 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 87.40% ±0.33 | 85.97% ±0.41 | 1.43 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 69.40% ±0.91 | 61.80% ±0.97 | 7.59 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 9.29% ±2.09 | 24.86% ±3.12 | 15.56 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 72.88% ±0.86 | 71.67% ±0.89 | 1.21 pts | Reported ±1 SE ranges overlap |
| MedCode | 59.06% ±2.00 | 43.29% ±2.31 | 15.77 pts | Reported ±1 SE ranges do not overlap |
| MedScribe | 76.11% ±1.92 | 86.88% ±1.94 | 10.77 pts | Reported ±1 SE ranges do not overlap |
| ProofBench v1.1 | 26.00% ±4.41 | 31.00% ±4.65 | 5.00 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 95.45% ±1.05 | 92.93% ±1.29 | 2.52 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 90.99% ±0.28 | 89.22% ±0.31 | 1.77 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 88.21% ±0.78 | 61.79% ±1.17 | 26.42 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 48.68% ±3.29 | 34.97% ±3.32 | 13.71 pts | Reported ±1 SE ranges do not overlap |
| IOI | 51.83% ±2.59 | 40.56% ±3.95 | 11.28 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 88.48% ±0.93 | 87.35% ±0.96 | 1.13 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 | 78.80% ±1.83 | 86.60% ±1.52 | 7.80 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 2.52% ±0.51 | 8.59% ±1.34 | 6.06 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 32.03% ±4.34 | 69.00% ±4.48 | 36.97 pts | Reported ±1 SE ranges do not overlap |
| SRE Bench | 0.00% ±0.00 | 0.76% ±0.54 | 0.76 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Gemini 3.1 Pro Preview (02/26) average | Grok 4.5 average |
|---|---|---|
| Legal | 54.04% | 61.97% |
| Finance | 50.52% | 52.78% |
| Healthcare | 67.59% | 65.09% |
| Math | 26.00% | 31.00% |
| Academic | 91.55% | 81.31% |
| Education | 48.68% | 34.97% |
| Coding | 42.28% | 48.68% |
| Cyber | 0.00% | 0.76% |
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 3.1 Pro Preview (02/26) cost | Grok 4.5 cost | Gemini 3.1 Pro Preview (02/26) latency | Grok 4.5 latency |
|---|---|---|---|---|
| Legal Research Bench | $1.10 | $1.09 | 4m02s | 15m54s |
| LegalBench | N/A | N/A | 10.06s | 67.88s |
| MortgageTax | N/A | N/A | 23.62s | 117.46s |
| Tax Agent Bench | $0.30 | $0.47 | 100.96s | 4m44s |
| TaxEval v2 | N/A | N/A | 41.93s | 3m15s |
| MedCode | N/A | N/A | 38.52s | 59.64s |
| MedScribe | N/A | N/A | 69.14s | 11m45s |
| ProofBench v1.1 | $0.78 | $0.57 | 14m49s | 11m09s |
| GPQA Diamond | N/A | N/A | 65.76s | 3m38s |
| MMLU Pro | N/A | N/A | 23.91s | 5m58s |
| MMMU Pro | N/A | N/A | 76.99s | 11m55s |
| SAGE | N/A | N/A | 61.17s | 2m50s |
| IOI | $1.25 | $5.64 | 12m03s | 1h27m |
| LiveCodeBench | N/A | N/A | 88.89s | 9m49s |
| ProgramBench | $2.63 | $4.12 | 12m36s | 18m59s |
| SWE-bench | $0.78 | $0.54 | 5m12s | 3m20s |
| Terminal-Bench 4.0 | $4.11 | $5.52 | 18m44s | 33m43s |
| Vibe Code Bench v1.1 | $3.83 | $4.10 | 20m12s | 13m50s |
| SRE Bench | $11.34 | $13.42 | 1h03m | 1h06m |
Results available only for Gemini 3.1 Pro Preview (02/26)
- Vals Science Index
- BioMysteryBench
- MysteryMechanism
- Terminal-Bench-Science
- Vibe Code Bench 1-100
- Public Benefits Bench v1.1
Results available only for Grok 4.5
- SkillsBench
- CyberBench v1.1
- Time Horizon Index: KSP