Head to head
GPT-5.4 vs GPT-5.3 Codex
On a blended workload of one million input tokens and three million output tokens, GPT-5.3 Codex is 1.1× cheaper than GPT-5.4.
| GPT-5.4 | GPT-5.3 Codex | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Input / MTok | $2.5 | $1.75 |
| Output / MTok | $15 | $14 |
| Cache read / MTok | $0.25 | $0.175 |
| Context window | — | — |
| Blended 1M in + 3M out | $47.50 | $43.75 |
| Status | ga | ga |