Price voice APIs with accumulated context, not audio rates alone

Compare voice API costs using whole-session usage and the provider's billing rules, rather than multiplying a headline audio rate by call duration. Gemini Live documents re-billing accumulated audio context on each turn; GPT-Live instead charges elapsed session time plus backend model and tool usage. A cheaper audio-token rate can therefore lose its apparent advantage as turns accumulate. Estimate the actual workload, including silence, prompt reuse, tools, and telephony, before choosing. These billing shapes do not establish which provider is cheaper for a particular call or which sounds more natural.

Applies when: Comparing native voice models with chained speech pipelines for multi-turn calls. Keywords: voice pricing Gemini Live OpenAI context audio tokens billing