Pricing & Cost

Output Tokens

Output tokens are tokens in the model's generated response. APIs typically charge 2-4x more per output token than input tokens. Output token cost scales with response length. Techniques reducing output: constraining max_tokens, better prompting (shorter answers), summarization. Output optimization is less impactful than input in many cases. Long-context applications with short responses prefer per-input pricing. Cost structure matters for application design.