GPT-5.5 API Pricing & Cost Calculator

How this is calculated

GPT-5.5 is priced at $5.00 per million input tokens and $30.00 per million output tokens, supporting a 90% caching discount ($0.50 per million) and a 50% batch discount ($2.50 per million inputs, $15.00 per million outputs).

Verdict

The headline rates are $5.00/M input and $30.00/M output, so a typical 100K input + 5K output call runs about $0.65 before discounts. The two cost levers pull that down fast: the 90% cached-input discount sets cache reads at $0.50/M. For a stable 3K-token prefix reused 100K times per day, billing all 300M prefix tokens as cache reads cuts that input component from $1,500 to $150 per day, excluding the initial uncached fill, other input, and output, and the 50% batch discount halves both rows ($2.50/M input, $15.00/M output) for overnight bulk runs. Those economics only make sense when the task genuinely needs GPT-5.5's frontier reasoning. For routine chat and RAG, route to a cheaper workhorse and reserve it for the hardest calls.