Gemini 3.5 Flash Lite vs GPT 5.4 Mini: Benchmark Comparison
Gemini 3.5 Flash Lite has the higher score on 9 of 16 shared benchmarks; GPT 5.4 Mini leads on 5.
The largest observed score gap is 10.81 pts on Vibe Code Bench v1.1 , where GPT 5.4 Mini leads.
Reported ±1 standard-error ranges overlap on 9 of 16 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 3.5 Flash Lite | GPT 5.4 Mini | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 13.94% ±2.41 | 12.50% ±2.30 | 1.44 pts | Reported ±1 SE ranges overlap |
| EMB | 43.23% ±3.10 | 45.43% ±3.65 | 2.20 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 47.44% ±0.51 | 45.36% ±0.45 | 2.08 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 68.68% ±0.91 | 63.51% ±0.91 | 5.17 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 37.45% ±3.12 | 36.59% ±3.10 | 0.87 pts | Reported ±1 SE ranges overlap |
| TaxEval v2 | 72.61% ±0.88 | 71.22% ±0.90 | 1.39 pts | Reported ±1 SE ranges overlap |
| GPQA Diamond | 83.84% ±1.85 | 83.08% ±2.46 | 0.76 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 85.84% ±0.34 | 84.55% ±0.36 | 1.28 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 83.58% ±0.89 | 79.25% ±0.97 | 4.34 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 47.30% ±3.38 | 50.81% ±3.40 | 3.51 pts | Reported ±1 SE ranges overlap |
| Code Migration | 6.15% ±1.71 | 12.94% ±3.64 | 6.79 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 79.01% ±1.10 | 81.47% ±1.09 | 2.45 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 75.00% ±1.94 | 73.00% ±1.99 | 2.00 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 37.16% ±4.63 | 47.97% ±5.61 | 10.81 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Gemini 3.5 Flash Lite average | GPT 5.4 Mini average |
|---|---|---|
| Legal | 6.97% | 6.25% |
| Finance | 53.88% | 52.42% |
| Academic | 84.42% | 82.29% |
| Education | 47.30% | 50.81% |
| Coding | 39.46% | 43.07% |
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 3.5 Flash Lite cost | GPT 5.4 Mini cost | Gemini 3.5 Flash Lite latency | GPT 5.4 Mini latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $0.33 | $1.00 | 3m53s | 11m21s |
| Legal Research Bench | $0.12 | $1.86 | 59.20s | 37m36s |
| EMB | $1.23 | $2.15 | 8m14s | 43m21s |
| Finance Agent (v2) | $0.39 | $1.20 | 2m23s | 35m54s |
| MortgageTax | N/A | N/A | 4.88s | 2m06s |
| Tax Agent Bench | $0.08 | $0.81 | 47.46s | 16m46s |
| TaxEval v2 | N/A | N/A | 8.81s | 44.58s |
| GPQA Diamond | N/A | N/A | 13.85s | 63.11s |
| MMLU Pro | N/A | N/A | 4.82s | 20.00s |
| MMMU Pro | N/A | N/A | 7.80s | 50.81s |
| SAGE | N/A | N/A | 10.75s | 2m21s |
| Code Migration | $0.53 | $1.95 | 22m02s | 28m53s |
| LiveCodeBench | N/A | N/A | 21.08s | 3m09s |
| ProgramBench | N/A | N/A | 7m15s | 53m45s |
| SWE-bench | $0.31 | $0.51 | 7m20s | 5m26s |
| Vibe Code Bench v1.1 | $0.83 | $1.19 | 8m22s | 34m15s |
Results available only for Gemini 3.5 Flash Lite
- LegalBench
- MedCode
- MedScribe
- CyberBench v1.1
- Public Benefits Bench v1.1
Results available only for GPT 5.4 Mini
- Vals Index
- Terminal-Bench 4.0