Reasoning models (o3, Gemini Thinking) bill hidden thinking tokens — see the real cost.
Reasoning models can spend thousands of invisible tokens per call. If you budget only for the visible answer, the bill blindsides you. Compare against standard models on the LLM token cost calculator.
Free reasoning-tokens cost calculator — visible output plus hidden reasoning tokens on o3-class models, per call and per month.
Yes. Models like o3 and Gemini Thinking generate hidden reasoning tokens before the visible answer, and those are billed at the output rate. They often outnumber the visible output, so the real cost is far higher than the answer length suggests.
Use a reasoning-effort setting where available, switch to a standard model for simple tasks, and measure actual reasoning-token usage — it varies wildly by prompt.