Pricing and Comparison
Prompt Caching
Cached input tokens are automatically discounted at 50% off the standard input price. Caching kicks in automatically for repeated prompt prefixes.All models support tool calling, structured outputs, web search, and text +
image input with text output.