Grok 4.5 vs MiniMax-M3: Benchmark Comparison
Grok 4.5 has the higher score on 12 of 18 shared benchmarks; MiniMax-M3 leads on 6.
The largest observed score gap is 21.44 pts on Vibe Code Bench v1.1 , where Grok 4.5 leads.
Reported ±1 standard-error ranges overlap on 7 of 18 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 | Grok 4.5 | MiniMax-M3 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 37.98% ±3.37 | 29.81% ±3.18 | 8.17 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 85.97% ±0.41 | 85.42% ±0.44 | 0.55 pts | Reported ±1 SE ranges overlap |
| MortgageTax | 61.80% ±0.97 | 68.36% ±0.91 | 6.56 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 24.86% ±3.12 | 20.59% ±2.87 | 4.27 pts | Reported ±1 SE ranges overlap |
| TaxEval v2 | 71.67% ±0.89 | 72.73% ±0.86 | 1.06 pts | Reported ±1 SE ranges overlap |
| MedCode | 43.29% ±2.31 | 46.29% ±2.10 | 3.00 pts | Reported ±1 SE ranges overlap |
| MedScribe | 86.88% ±1.94 | 87.25% ±1.96 | 0.37 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 31.00% ±4.65 | 18.00% ±3.86 | 13.00 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 92.93% ±1.29 | 92.68% ±1.44 | 0.25 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 89.22% ±0.31 | 84.22% ±0.36 | 5.00 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 61.79% ±1.17 | 81.16% ±0.94 | 19.36 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 34.97% ±3.32 | 50.57% ±3.44 | 15.61 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 87.35% ±0.96 | 82.15% ±1.05 | 5.20 pts | Reported ±1 SE ranges do not overlap |
| SkillsBench | 66.03% ±4.60 | 51.50% ±4.50 | 14.54 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench | 86.60% ±1.52 | 75.00% ±1.94 | 11.60 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench 4.0 | 8.59% ±1.34 | 1.01% ±1.01 | 7.58 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 69.00% ±4.48 | 47.57% ±5.44 | 21.44 pts | Reported ±1 SE ranges do not overlap |
| CyberBench v1.1 | 71.01% ±5.79 | 65.00% ±6.10 | 6.01 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Grok 4.5 average | MiniMax-M3 average |
|---|---|---|
| Legal | 61.97% | 57.61% |
| Finance | 52.78% | 53.89% |
| Healthcare | 65.09% | 66.77% |
| Math | 31.00% | 18.00% |
| Academic | 81.31% | 86.02% |
| Education | 34.97% | 50.57% |
| Coding | 63.51% | 51.45% |
| Cyber | 71.01% | 65.00% |
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 | Grok 4.5 cost | MiniMax-M3 cost | Grok 4.5 latency | MiniMax-M3 latency |
|---|---|---|---|---|
| Legal Research Bench | $1.09 | $0.34 | 15m54s | 13m34s |
| LegalBench | N/A | N/A | 67.88s | 8.21s |
| MortgageTax | N/A | N/A | 117.46s | 26.76s |
| Tax Agent Bench | $0.47 | $0.16 | 4m44s | 5m15s |
| TaxEval v2 | N/A | N/A | 3m15s | 94.04s |
| MedCode | N/A | N/A | 59.64s | 63.12s |
| MedScribe | N/A | N/A | 11m45s | 2m04s |
| ProofBench v1.1 | $0.57 | $0.42 | 11m09s | 11m59s |
| GPQA Diamond | N/A | N/A | 3m38s | 4m39s |
| MMLU Pro | N/A | N/A | 5m58s | 41.39s |
| MMMU Pro | N/A | N/A | 11m55s | 68.35s |
| SAGE | N/A | N/A | 2m50s | 118.54s |
| LiveCodeBench | N/A | N/A | 9m49s | 6m07s |
| SkillsBench | $0.58 | $0.50 | 2m16s | 15m28s |
| SWE-bench | $0.54 | $0.42 | 3m20s | 12m06s |
| Terminal-Bench 4.0 | $5.52 | $7.23 | 33m43s | 1h16m |
| Vibe Code Bench v1.1 | $4.10 | $6.45 | 13m50s | 1h25m |
| CyberBench v1.1 | $2.95 | $3.41 | 14m08s | 40m56s |
Results available only for Grok 4.5
- IOI
- ProgramBench
- SRE Bench
- Time Horizon Index: KSP
Results available only for MiniMax-M3
- Vals Finance Index
- Vals Inference Provider Index
- EMB v1.1
- Finance Agent v2.1
- Code Migration v1.1
- SWE-bench Multimodal
- Vibe Code Bench 1-100
- Public Benefits Bench v1.1