Head to head
GPT-5.4 nano vs GPT-5.2
On a blended workload of one million input tokens and three million output tokens, GPT-5.4 nano is 11.1× cheaper than GPT-5.2.
| GPT-5.4 nano | GPT-5.2 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Input / MTok | $0.2 | $1.75 |
| Output / MTok | $1.25 | $14 |
| Cache read / MTok | $0.02 | $0.175 |
| Context window | — | — |
| Blended 1M in + 3M out | $3.95 | $43.75 |
| Status | ga | ga |