Qwen 3.7 Max vs GPT 5.4 Mini: Benchmark Comparison
Qwen 3.7 Max has the higher score on 10 of 12 shared benchmarks; GPT 5.4 Mini leads on 2.
The largest observed score gap is 12.98 pts on Legal Research Bench , where Qwen 3.7 Max leads.
Reported ±1 standard-error ranges overlap on 2 of 12 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 | Qwen 3.7 Max | GPT 5.4 Mini | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 1.67% ±0.83 | 0.00% ±0.00 | 1.67 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 25.48% ±3.03 | 12.50% ±2.30 | 12.98 pts | Reported ±1 SE ranges do not overlap |
| EMB | 57.04% ±2.85 | 45.43% ±3.65 | 11.61 pts | Reported ±1 SE ranges do not overlap |
| Finance Agent (v2) | 47.78% ±0.73 | 45.36% ±0.45 | 2.42 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 47.41% ±3.24 | 36.59% ±3.10 | 10.82 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 75.31% ±0.84 | 71.22% ±0.90 | 4.09 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 90.15% ±1.50 | 83.08% ±2.46 | 7.07 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 89.31% ±0.30 | 84.55% ±0.36 | 4.76 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 13.07% ±2.86 | 12.94% ±3.64 | 0.13 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 87.06% ±0.97 | 81.47% ±1.09 | 5.59 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench | 68.80% ±2.07 | 73.00% ±1.99 | 4.20 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 47.67% ±4.63 | 47.97% ±5.61 | 0.30 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Qwen 3.7 Max average | GPT 5.4 Mini average |
|---|---|---|
| Legal | 13.57% | 6.25% |
| Finance | 56.88% | 49.65% |
| Academic | 89.73% | 83.82% |
| Coding | 54.15% | 53.84% |
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 | Qwen 3.7 Max cost | GPT 5.4 Mini cost | Qwen 3.7 Max latency | GPT 5.4 Mini latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $1.58 | $1.00 | 32m03s | 11m21s |
| Legal Research Bench | $1.40 | $1.86 | 13m29s | 37m36s |
| EMB | $2.86 | $2.15 | 32m37s | 43m21s |
| Finance Agent (v2) | $0.55 | $1.20 | 10m35s | 35m54s |
| Tax Agent Bench | $0.44 | $0.81 | 3m30s | 16m46s |
| TaxEval v2 | N/A | N/A | 57.68s | 44.58s |
| GPQA Diamond | N/A | N/A | 2m07s | 63.11s |
| MMLU Pro | N/A | N/A | 29.15s | 20.00s |
| Code Migration | $1.93 | $1.95 | 25m32s | 28m53s |
| LiveCodeBench | N/A | N/A | 3m12s | 3m09s |
| SWE-bench | $3.23 | $0.51 | 16m16s | 5m26s |
| Vibe Code Bench v1.1 | $11.40 | $1.19 | 1h18m | 34m15s |
Results available only for Qwen 3.7 Max
- LegalBench
- MedCode
- MedScribe
Results available only for GPT 5.4 Mini
- Vals Index
- MortgageTax
- MMMU Pro
- SAGE
- ProgramBench
- Terminal-Bench 4.0