Estimate your monthly AI API costs across different providers and models. Adjust your usage pattern below.
AI APIs charge per token — roughly 4 characters or ¾ of a word. You pay separately for input tokens (what you send) and output tokens (what the model generates). Prices vary dramatically between models and providers.
For an embedded AI project like the ProjectOSAI Companion Watch, a typical voice conversation uses about 200-500 input tokens and 100-200 output tokens. At 20 conversations per day, most models cost under $15/month.
Prices shown are based on publicly listed API rates and may change. Always check the provider's current pricing page for the latest rates.