InceptionMercury
Mercury Voice
Mercury Voice supports tool calls and long instructions within voice-agent systems. Enterprise customers must request API activation.
Restricted access
What the release provides
The model connects to the language-model stage of LiveKit, Pipecat, Vapi, Retell or a custom voice stack. It is supplied through an OpenAI-compatible endpoint.Inception: Introducing Mercury Voice
How to use it well
Practical guidance. Keep speech recognition, language generation and speech synthesis as separate entries in your latency measurements. Check interruption handling with the full voice application.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Restricted access |
| Access | API |
| Model identifiers | mercury-voice |
| Context window | 128,000 tokens Advertised 128K context. Inception Platform model specifications |
| Maximum output | 50,000 tokens API output limit. Inception Platform model specifications |
| Input price | 0.2 USD per million tokens Launch discount. Inception: Introducing Mercury Voice |
| Output price | 0.75 USD per million tokens Launch discount. Inception: Introducing Mercury Voice |
| Input and output | Text; speech processing is supplied by the voice stack.Inception Platform model specifications |