Amazon Nova 2 Omni
Preview reasoning model that takes text, image, video and speech in and generates text and images; 1M context.
One model for multimodal input and mixed output (text plus image generation/editing, speech transcription and summarisation), described by AWS as the first reasoning model with all four input types plus image output. Early access limited to Nova Forge customers.
- Date
- Tuesday, 2 December 2025
- Lab
- Amazon
- Kind
- model
- Access
- research preview
Figures
| Measure | Value | Measured by |
|---|---|---|
| Context window | 1M tokens 200+ languages for text, 10 for speech | company |
'Industry's first' is an Amazon claim.
Sources
This record was checked against its sources on 6 October 2026. How we check