← Back to Voice models
Amazon (Nova)
Amazon Nova 2 Sonic
Released Dec 2, 2025
Amazon's second-generation speech-to-speech model on Bedrock for real-time voice agents, with native telephony integrations. Nova 2.5 Sonic has since superseded it.
- Price per minute
- —
- Pricing
- Token-priced on Bedrock (speech and text input/output); exact rates not retrievable from aws.amazon.com/bedrock/pricing.
- Inputs
- audio, text
- Tool calling
- Yes
- Languages
- 7
- Voices
- —
- Big Bench Audio
- 88.1%
- Access
- Generally available
- Licence
- Proprietary
Strengths & weaknesses
Strengths
- Integrates with Amazon Connect, Twilio, Vonage, AudioCodes, LiveKit and Pipecat
- Asynchronous tool handling and RAG grounding
- 1M-token context window
Weaknesses
- 8-minute connection limit, so long calls need session renewal
- Superseded by Nova 2.5 Sonic
- Big Bench Audio 88.1% (AA, Mar 2026 version)
Not notable (AA Big Bench Audio #21). The Bedrock model card says Active, with EOL no sooner than 2026-12-02; drop it if superseded models are out of scope. AA's listing is 'Nova 2.0 Sonic (Mar 2026)', which suggests a March 2026 update; no AWS page for that update was found. Voice count conflicts: the language-support table lists 16 voice IDs, and versions of the service card say 18 and 22, so it's null. Pricing rows on the Bedrock page didn't render.
Links
API model strings
- bedrock
amazon.nova-2-sonic-v1:0