GPT 5.4 Mini vs MiMo V2.5: Benchmark Comparison
GPT 5.4 Mini has the higher score on 9 of 15 shared benchmarks; MiMo V2.5 leads on 6.
The largest observed score gap is 9.67 pts on EMB , where MiMo V2.5 leads.
Reported ±1 standard-error ranges overlap on 8 of 15 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 | GPT 5.4 Mini | MiMo V2.5 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 0.00% ±0.00 | 1.67% ±0.00 | 1.67 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 12.50% ±2.30 | 9.13% ±2.00 | 3.37 pts | Reported ±1 SE ranges overlap |
| EMB | 45.43% ±3.65 | 55.09% ±3.13 | 9.67 pts | Reported ±1 SE ranges do not overlap |
| Finance Agent (v2) | 45.36% ±0.45 | 36.73% ±0.11 | 8.63 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 63.51% ±0.91 | 59.26% ±0.98 | 4.25 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 36.59% ±3.10 | 29.16% ±2.71 | 7.42 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 71.22% ±0.90 | 71.83% ±0.88 | 0.61 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 83.08% ±2.46 | 81.57% ±2.05 | 1.51 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 84.55% ±0.36 | 82.93% ±0.37 | 1.62 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 79.25% ±0.97 | 80.00% ±0.96 | 0.75 pts | Reported ±1 SE ranges overlap |
| SAGE | 50.81% ±3.40 | 43.27% ±3.39 | 7.55 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 12.94% ±3.64 | 14.23% ±3.69 | 1.29 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 81.47% ±1.09 | 81.51% ±1.07 | 0.05 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 73.00% ±1.99 | 71.00% ±2.03 | 2.00 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 47.97% ±5.61 | 42.17% ±4.57 | 5.80 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | GPT 5.4 Mini average | MiMo V2.5 average |
|---|---|---|
| Legal | 6.25% | 5.40% |
| Finance | 52.42% | 50.41% |
| Academic | 82.29% | 81.50% |
| Education | 50.81% | 43.27% |
| Coding | 53.84% | 52.23% |
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 | GPT 5.4 Mini cost | MiMo V2.5 cost | GPT 5.4 Mini latency | MiMo V2.5 latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $1.00 | $0.04 | 11m21s | 6m59s |
| Legal Research Bench | $1.86 | $0.03 | 37m36s | 6m31s |
| EMB | $2.15 | $0.08 | 43m21s | 30m38s |
| Finance Agent (v2) | $1.20 | $0.09 | 35m54s | 5m39s |
| MortgageTax | N/A | N/A | 2m06s | 32.94s |
| Tax Agent Bench | $0.81 | $0.02 | 16m46s | 8m21s |
| TaxEval v2 | N/A | N/A | 44.58s | 26.30s |
| GPQA Diamond | N/A | N/A | 63.11s | 89.39s |
| MMLU Pro | N/A | N/A | 20.00s | 30.25s |
| MMMU Pro | N/A | N/A | 50.81s | 42.47s |
| SAGE | N/A | N/A | 2m21s | 52.66s |
| Code Migration | $1.95 | $0.10 | 28m53s | 37m55s |
| LiveCodeBench | N/A | N/A | 3m09s | 108.61s |
| SWE-bench | $0.51 | $0.01 | 5m26s | 4m04s |
| Vibe Code Bench v1.1 | $1.19 | $0.07 | 34m15s | 26m20s |
Results available only for GPT 5.4 Mini
- Vals Index
- ProgramBench
- Terminal-Bench 4.0
Results available only for MiMo V2.5
- LegalBench
- MedCode
- MedScribe
- ProofBench v1.1