Atlas

Models

← All models

Nova 2 Omni (Preview)

Amazon Web Services (AWS) · 2025-12-02

Amazon Nova 2 Omni (Preview) is a unified multimodal reasoning model in Amazon's second-generation Nova family that accepts text, image, video, and speech inputs and can generate both text and images. Announced at AWS re:Invent 2025 with a one-million-token context window, it combines multimodal understanding, speech transcription, image generation and editing, and flexible reasoning controls in a single model.

Benchmark scores

BenchmarkScore
AA-LCR53.7
AA-Omniscience Index-50.0
AIME 202589.7
Artificial Analysis Intelligence Index20.9
Artificial Analysis Omniscience Accuracy18.5
Artificial Analysis Omniscience Hallucination Rate84.0
Capability138.2
CritPt0.0
Global MMLU Lite81.0
GPQA Diamond76.0
Humanity's Last Exam (Text-Only)6.8
IFBench (Artificial Analysis)66.2
LiveCodeBench (Artificial Analysis)66.0
MMLU Pro80.9
MMMU Pro61.9
SciCode (Artificial Analysis)36.2
Terminal-Bench Hard6.8
τ²-bench (Telecom)80.4
Loading Atlas data…