ElevenLabs launches Eleven Realtime voice models at $0.08 a minute plus tokens

ElevenLabs released Eleven Realtime, a pair of speech-to-speech models for voice agents, on October 20, 2026. Eleven Realtime v1 Mini and v1 Max both cost $0.08 per minute of connected audio, plus separate charges for input and output tokens. The company says the models work in more than 90 languages and with any of the 17,000+ voices in its library.

What’s new

ElevenLabs calls Eleven Realtime “a new class of Speech to Speech models” and offers it in two tiers:

• v1 Mini (eleven_realtime_v1_mini): for natural conversations at low cost, such as AI assistants, teachers and interactive characters.
• v1 Max: for harder tasks such as customer support, lead qualification and multi-step workflows, with tool calling for lookups and actions.

There are two ways to use them. In ElevenAgents, you pick a Realtime model from the model dropdown and keep the platform’s phone, web and WhatsApp channels, knowledge base, guardrails and monitoring. Or you call the Realtime API directly, which ElevenLabs says is compatible with OpenAI’s Realtime API: you change the endpoint and model name in existing code. The API connects over WebSocket only, so WebRTC setups must move to WebSocket first, and it doesn’t include the ElevenAgents platform features.

What it costs

Per unit	Realtime v1 Mini	Realtime v1 Max
Connected audio, per minute	$0.08	$0.08
Input tokens, per 1M	$0.20	$1.80
Output tokens, per 1M	$6.00	$9.00
Burst audio above concurrency	$0.16 per minute	$0.16 per minute

The API pricing page also lists six rows of included minutes, from about 14 to about 11,794 for Mini and about 12 to about 9,714 for Max. It says these are estimates based on typical token usage and that actual minutes vary with conversation length and tool use. The page doesn’t label those rows by plan and doesn’t list a free-tier allowance for either model. Burst pricing applies to audio beyond your plan’s limit on concurrent calls, which the page lists as 4 to 40.

What to do

If you already run a voice agent on OpenAI’s Realtime API, the compatibility claim makes a trial cheap to set up, but budget for both the per-minute audio charge and tokens. Token use, not minutes, will decide the gap between Mini and Max. Compare other voice tools on our AI voice page.