Amazon Nova 2.5 Sonic
Amazon's generally available real-time speech-to-speech model in Bedrock, with a 256K context window and asynchronous tool calling.
Overview
AWS announced general availability of Amazon Nova 2.5 Sonic on October 5, 2026, a real-time speech-to-speech model for voice agents. It improves reasoning, instruction following and tool-calling accuracy over Nova 2 Sonic, supports asynchronous tool calling, a 256K context window, and voice plus text in one session. It runs in Amazon Bedrock in US East (N. Virginia), US West (Oregon), Europe (Stockholm) and Asia Pacific (Tokyo), and works with Strands Bidi Agents, also now generally available.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Key features
- 256K context window
- Asynchronous tool calling during a live conversation
- Voice and text in the same session
- Controllable turn-taking and speech understanding
- Expressive voices in seven languages
- Bedrock regions: us-east-1, us-west-2, eu-north-1, ap-northeast-1
Use cases
- Contact-center voice agents on Amazon Bedrock
- Voice assistants that call backend tools mid-conversation
- Multi-step spoken workflows built with Strands Bidi Agents
Pricing
Same pricing as Nova 2 Sonic; AWS did not publish rates in the announcement.
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- 5 Oct 2026
- Record updated
- 15 Oct 2026
- Record last checked
- 15 Oct 2026