MiniMax-M3 vs GPT 5.4 Mini: Benchmark Comparison
MiniMax-M3 has the higher score on 13 of 17 shared benchmarks; GPT 5.4 Mini leads on 4.
The largest observed score gap is 17.31 pts on Legal Research Bench , where MiniMax-M3 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 | MiniMax-M3 | GPT 5.4 Mini | Gap | Reported uncertainty |
|---|---|---|---|---|
| Vals Index | 36.53% ±1.17 | 33.17% ±1.28 | 3.37 pts | Reported ±1 SE ranges do not overlap |
| Harvey's Legal Agent Benchmark | 4.17% ±1.65 | 0.00% ±0.00 | 4.17 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 29.81% ±3.18 | 12.50% ±2.30 | 17.31 pts | Reported ±1 SE ranges do not overlap |
| EMB | 47.76% ±2.95 | 45.43% ±3.65 | 2.33 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 48.27% ±0.44 | 45.36% ±0.45 | 2.91 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 68.36% ±0.91 | 63.51% ±0.91 | 4.85 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 49.69% ±3.25 | 36.59% ±3.10 | 13.11 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 72.73% ±0.86 | 71.22% ±0.90 | 1.51 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 92.68% ±1.44 | 83.08% ±2.46 | 9.60 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 84.22% ±0.36 | 84.55% ±0.36 | 0.33 pts | Reported ±1 SE ranges overlap |
| MMMU Pro | 81.16% ±0.94 | 79.25% ±0.97 | 1.91 pts | Reported ±1 SE ranges overlap |
| SAGE | 50.57% ±3.44 | 50.81% ±3.40 | 0.24 pts | Reported ±1 SE ranges overlap |
| Code Migration | 19.93% ±3.94 | 12.94% ±3.64 | 6.99 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 82.15% ±1.05 | 81.47% ±1.09 | 0.69 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 75.00% ±1.94 | 73.00% ±1.99 | 2.00 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 1.01% ±1.01 | 2.52% ±1.01 | 1.51 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 47.57% ±5.44 | 47.97% ±5.61 | 0.40 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | MiniMax-M3 average | GPT 5.4 Mini average |
|---|---|---|
| Legal | 16.99% | 6.25% |
| Finance | 57.36% | 52.42% |
| Academic | 86.02% | 82.29% |
| Education | 50.57% | 50.81% |
| Coding | 45.13% | 43.58% |
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 | MiniMax-M3 cost | GPT 5.4 Mini cost | MiniMax-M3 latency | GPT 5.4 Mini latency |
|---|---|---|---|---|
| Vals Index | $3.08 | $1.45 | 38m55s | 29m49s |
| Harvey's Legal Agent Benchmark | $1.46 | $1.00 | 22m31s | 11m21s |
| Legal Research Bench | $0.34 | $1.86 | 13m34s | 37m36s |
| EMB | $2.09 | $2.15 | 31m30s | 43m21s |
| Finance Agent (v2) | $0.32 | $1.20 | 8m17s | 35m54s |
| MortgageTax | N/A | N/A | 26.76s | 2m06s |
| Tax Agent Bench | $0.16 | $0.81 | 5m15s | 16m46s |
| TaxEval v2 | N/A | N/A | 94.04s | 44.58s |
| GPQA Diamond | N/A | N/A | 4m39s | 63.11s |
| MMLU Pro | N/A | N/A | 41.39s | 20.00s |
| MMMU Pro | N/A | N/A | 68.35s | 50.81s |
| SAGE | N/A | N/A | 118.54s | 2m21s |
| Code Migration | $7.07 | $1.95 | 1h14m | 28m53s |
| LiveCodeBench | N/A | N/A | 6m07s | 3m09s |
| SWE-bench | $0.42 | $0.51 | 12m06s | 5m26s |
| Terminal-Bench 4.0 | $7.23 | $1.43 | 1h16m | 29m41s |
| Vibe Code Bench v1.1 | $6.45 | $1.19 | 1h25m | 34m15s |
Results available only for MiniMax-M3
- LegalBench
- MedCode
- MedScribe
- ProofBench v1.1
- SkillsBench
- Vibe Code Bench 1-100
- CyberBench v1.1
- Public Benefits Bench v1.1
Results available only for GPT 5.4 Mini
- ProgramBench