Side by side pricing and context for GPT-5 Mini and Claude Sonnet 4.6. Rates are curated catalog values with last verified dates on each model page.
GPT-5 Mini
Lower-cost GPT-5 tier for high-volume chat and light reasoning. Good default when GPT-5 is overkill but you still want the GPT-5 family tokenizer (Exact in TokenCALC).
Input $0.250 / Output $2.00 per 1M
Context 128K · Exact (o200k_base)
Claude Sonnet 4.6
Previous Sonnet generation still useful for stable production prompts. Compare against Sonnet 5 on your real traffic mix before migrating.
Input $3.00 / Output $15.00 per 1M
Context 1M · Approx (Anthropic family)
Quick take
- GPT-5 Mini is cheaper on input in this catalog snapshot.
- GPT-5 Mini is cheaper on output in this catalog snapshot.
- Both last verified on their model pages (2026-08-02 / 2026-08-02).
Rate table
| Metric | GPT-5 Mini | Claude Sonnet 4.6 |
|---|---|---|
| Input / 1M | $0.250 | $3.00 |
| Output / 1M | $2.00 | $15.00 |
| Cached input / 1M | $0.025 | $0.300 |
| Context | 128K | 1M |
| Tokenizer | Exact (o200k_base) | Approx (Anthropic family) |