Head to head
Claude Sonnet 5 vs GPT-5.3 Codex
For tasks, about a month.
No verdict
Cost per solved task needs a published pass rate for both models. At least one has none, so it is reported as missing rather than estimated.
| Figure | Claude Sonnet 5 | GPT-5.3 Codex |
|---|---|---|
| Input $/Mtok | $2 | $1.75 |
| Output $/Mtok | $10 | $14 |
| Cached input $/Mtok | $0.2 | $0.175 |
| Pass rate | missing | missing |
| Cost per solved task | missing | missing |
| Cost per month | missing | missing |
Cost per month against tasks per month · drag the marker
No line to draw: neither model has a published pass rate.
verified 2026-08-21 · prices from each provider's pricing page; measured rows ignore the tier · Methodology · All models at these settings