Gemini 3.5 Flash Lite vs GPT 5.4 Nano: Benchmark Comparison
Gemini 3.5 Flash Lite has the higher score on 13 of 18 shared benchmarks; GPT 5.4 Nano leads on 4.
The largest observed score gap is 11.06 pts on Vibe Code Bench v1.1 , where Gemini 3.5 Flash Lite leads.
Reported ±1 standard-error ranges overlap on 3 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 | Gemini 3.5 Flash Lite | GPT 5.4 Nano | 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 | 6.25% ±1.68 | 7.69 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 84.07% ±0.44 | 77.92% ±0.43 | 6.15 pts | Reported ±1 SE ranges do not overlap |
| EMB | 43.23% ±3.10 | 44.75% ±3.12 | 1.52 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 47.44% ±0.51 | 38.22% ±1.19 | 9.23 pts | Reported ±1 SE ranges do not overlap |
| MortgageTax | 68.68% ±0.91 | 59.10% ±0.96 | 9.58 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 37.45% ±3.12 | 26.58% ±2.52 | 10.87 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 72.61% ±0.88 | 67.42% ±0.93 | 5.19 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 43.49% ±1.95 | 41.03% ±2.26 | 2.46 pts | Reported ±1 SE ranges overlap |
| MedScribe | 70.89% ±2.03 | 77.09% ±1.89 | 6.20 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 83.84% ±1.85 | 77.53% ±2.42 | 6.31 pts | Reported ±1 SE ranges do not overlap |
| MMLU Pro | 85.84% ±0.34 | 77.17% ±0.43 | 8.66 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 83.58% ±0.89 | 73.58% ±1.06 | 10.00 pts | Reported ±1 SE ranges do not overlap |
| SAGE | 47.30% ±3.38 | 38.08% ±3.09 | 9.22 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 6.15% ±1.71 | 14.47% ±4.03 | 8.32 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 79.01% ±1.10 | 84.01% ±1.04 | 5.00 pts | Reported ±1 SE ranges do not overlap |
| SWE-bench | 75.00% ±1.94 | 69.80% ±2.06 | 5.20 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 37.16% ±4.63 | 26.10% ±5.08 | 11.06 pts | Reported ±1 SE ranges do not overlap |
Performance by category
| Category | Gemini 3.5 Flash Lite average | GPT 5.4 Nano average |
|---|---|---|
| Legal | 32.67% | 28.06% |
| Finance | 53.88% | 47.21% |
| Healthcare | 57.19% | 59.06% |
| Academic | 84.42% | 76.09% |
| Education | 47.30% | 38.08% |
| Coding | 49.33% | 48.59% |
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 Nano cost | Gemini 3.5 Flash Lite latency | GPT 5.4 Nano latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $0.33 | $0.18 | 3m53s | 6m03s |
| Legal Research Bench | $0.12 | $0.14 | 59.20s | 8m47s |
| LegalBench | N/A | N/A | 1.67s | 2.70s |
| EMB | $1.23 | $0.57 | 8m14s | 29m27s |
| Finance Agent (v2) | $0.39 | $0.16 | 2m23s | 5m35s |
| MortgageTax | N/A | N/A | 4.88s | 14.53s |
| Tax Agent Bench | $0.08 | $0.07 | 47.46s | 3m40s |
| TaxEval v2 | N/A | N/A | 8.81s | 15.90s |
| MedCode | N/A | N/A | 6.02s | 8.04s |
| MedScribe | N/A | N/A | 18.99s | 20.53s |
| GPQA Diamond | N/A | N/A | 13.85s | 29.50s |
| MMLU Pro | N/A | N/A | 4.82s | 8.21s |
| MMMU Pro | N/A | N/A | 7.80s | 21.91s |
| SAGE | N/A | N/A | 10.75s | 34.33s |
| Code Migration | $0.53 | $0.42 | 22m02s | 26m34s |
| LiveCodeBench | N/A | N/A | 21.08s | 60.09s |
| SWE-bench | $0.31 | $0.10 | 7m20s | 4m23s |
| Vibe Code Bench v1.1 | $0.83 | $1.28 | 8m22s | 54m36s |
Results available only for Gemini 3.5 Flash Lite
- ProgramBench
- CyberBench v1.1
- Public Benefits Bench v1.1
Results available only for GPT 5.4 Nano
None.