← All comparisons
GPT-5.4 nano (medium) vs GPT-5.2 (medium)
OpenAI vs OpenAI — side-by-side benchmark comparison
| GPT-5.4 nano (medium) | GPT-5.2 (medium) | |
|---|---|---|
| Intelligence Index | 38.1 | 46.6 |
| Coding Index | 35.0 | 44.2 |
| Math Index | — | 96.7 |
| Output speed (tok/s) | 152.6 | 0.0 |
| Blended price ($/1M) | $0.46 | $4.81 |
| Time to first token (s) | 2.09s | 0.00s |
| aime | — | — |
| aime 25 | — | 96.7% |
| artificial analysis coding index | 35.00 | 44.20 |
| artificial analysis intelligence index | 38.10 | 46.60 |
| artificial analysis math index | — | 96.70 |
| gpqa | 76.1% | 86.4% |
| hle | 14.7% | 24.9% |
| ifbench | 64.4% | 65.2% |
| lcr | 57.3% | 63.3% |
| livecodebench | — | 89.4% |
| math 500 | — | — |
| mmlu pro | — | 85.9% |
| scicode | 38.4% | 46.2% |
| tau2 | 52.6% | 74.3% |
| terminalbench hard | 33.3% | 43.2% |
Benchmark data from Artificial Analysis.