NVIDIA Unveils DeepSeek R1 Rival

AI just took a leap forward—smarter, faster, and more creative than ever. NVIDIA’s Llama3.1 Nemotron Ultra packs 25.3 billion parameters into a high-efficiency model that outperforms much larger rivals like Llama4Behemoth, delivering 4x the throughput of DeepSeek R1 and now available open-source on HuggingFace. Amazon’s Nova Sonic changes the voice AI game by combining speech understanding and generation into a single system, capturing tone, pace, and style for more human-like, emotionally aware responses across industries. And for creators, Nova Reel 1.1 enables multi-shot video generation up to 2 minutes from text and images, offering both automation and control for polished, on-brand content. Big moves, lean models—the AI evolution continues.

NVIDIA Llama3.1 Nemotron Ultra: Redefining AI Efficiency

NVIDIA unveiled the Llama3.1 Nemotron Ultra 253B, an open-source AI model with 25.3 billion parameters, optimized from Llama-3.1-405B. Despite its smaller size, it outperforms larger models like Meta's Llama4Behemoth and Maverick in benchmarks such as GPQA-Diamond and AIME2024/25, achieving four times the inference throughput of DeepSeek R1. Available on HuggingFace under a commercially friendly license, this model exemplifies efficiency and sustainability in AI design, challenging the industry's focus on parameter size. Its versatility spans code generation, scientific research, and natural language processing, offering developers a high-performance platform and businesses cost-effective deployment opportunities to drive innovation across sectors.

Amazon Nova Sonic: AI That Understands How You Say It

Amazon has introduced Nova Sonic, a groundbreaking AI model that integrates speech understanding and generation into a single system, enabling more natural and human-like voice interactions. This unified approach simplifies the development of voice applications by eliminating the need for multiple models, preserving crucial acoustic nuances like tone, prosody, and speaking style. Nova Sonic can adapt its responses to match the user's tone, style, and pace, and it handles natural pauses and interruptions gracefully. Available through Amazon Bedrock, the model supports a wide range of applications, including customer service automation, travel, education, and healthcare. It also generates text transcripts of user speech, allowing developers to integrate with other tools and APIs for enhanced functionality. Nova Sonic is designed to be cost-efficient and competitive with leading voice models from OpenAI and Google, offering low latency and industry-leading price performance.

Amazon Nova Reel 1.1: Create Multi-Shot Videos Up to 2 Minutes

Amazon Nova Reel 1.1 is an updated video generation model capable of creating multi-shot videos up to 2 minutes long from text descriptions and optional reference images. This enhancement allows for consistent style across shots, offering automated and manual modes for video creation through Amazon Bedrock. The automated mode uses a single prompt for the entire video, while the manual mode allows for custom prompts for each shot, providing greater creative control. Amazon Nova Reel 1.1 aims to improve creative productivity and reduce video production costs, making it suitable for marketing campaigns, product designs, and social media content. It is available in the US East (N. Virginia) AWS Region and can be accessed via the Amazon Bedrock console, AWS SDK, or AWS CLI, with pay-as-you-go pricing.

Hand Picked Video

In this video, I’ll show you how my AI-powered agent automatically finds and analyzes competitors using Firecrawl, Perplexity Sonar, and OpenAI APIs.

Top AI Products from this week

• EZsite AI - EZsite AI lets anyone build professional websites/apps with auto-saving to your database, capture leads with builtin AI chatbots, and boost sales using AI automations – no coding needed.

• Agno - Agno is a lightweight library for building Reasoning Agents with memory, knowledge, tools and native multi-modal support.

• chatWise - chatWise is the easiest way to monetize your expertise. You simply upload your content and the app instantly creates an AI-powered chatbot that others can subscribe to and learn from.

• Clarisign - World’s first AI that drafts, negotiates, and finalizes your contracts—so you can sign smarter and safer. It highlights risks, auto-fills key details, and cuts hours of paperwork down to minutes.

• taatoo - If you post it, they’ll steal it Unless you use taatoo... Protect your images with invisible watermarks powered by AI. Keep your rights without ruining your visuals. Stay safe out there

• Drooid - Drooid provides quick, unbiased news summaries by pulling information from multiple sources. The idea is to show different perspectives so you can get a complete picture of what’s happening, with no bias and no fluff.

This week in AI

• Gemini Live Update - Share your phone's camera or screen with Gemini Live for real-time AI assistance, available on Pixel 9 and Samsung Galaxy S25 devices127.

• IBM z17 Mainframe - IBM's z17 boosts AI capabilities with the Telum II processor, processing 450 billion inference operations daily.

• Genies AI Avatars - Genies introduces user-generated content tools for creating custom AI avatars, fostering limitless creativity and interoperability across experiences.

• Dire Wolf Revival - Colossal Biosciences uses genetic engineering to revive dire wolves, with pups Romulus and Remus born via surrogate mothers.

• Adobe Premiere Pro Update - Adobe launches Generative Extend, an AI tool extending video clips up to 2 seconds in 4K, alongside AI-powered search and caption translation features.