Paste your prompt and expected output. Pick a model. See the cost before you commit.
Token estimates are approximate (~+/-15% vs real tokenizers). Pricing from public API rates.