Post by DeepInfra

We're an official launch partner for NVIDIA Nemotron 3 Nano Omni — live today on DeepInfra. Nano Omni is the first multimodal model in the Nemotron 3 family: a single open model that understands images, video, audio, documents, and text in one unified inference pass. Built for agentic AI, it powers always-on sub-agents across use cases like computer use, document intelligence, and multimodal workflows—without stitching together separate models. Under the hood: ●  Unified vision and audio encoders—no separate perception models ●  30B-A3B hybrid MoE architecture with Mamba-transformer backbone ●  Long-context multimodal reasoning ●  Fully open weights, datasets, and recipes Available now on DeepInfra through our OpenAI-compatible API—with zero setup, low latency, and usage-based pricing. Thanks to the NVIDIA Nemotron team for the partnership. Try it today: https://lnkd.in/d2t5maJv #GenerativeAI #NVIDIANemotron #OpenSource #MultimodalAI #AgenticAI #OpenSource #Inference