Amazon AI Video Model Nova Reel Upgrade: Now Generates Videos Up to Two Minutes Long

AIbase基地

Published inAI News · 4 min read · Apr 8, 2025

Amazon recently upgraded its AI video generation model, Nova Reel, launching Nova Reel 1.1. This new version can generate videos up to two minutes long and create multi-shot videos with consistent stylistic elements across shots. First released in December 2024, Nova Reel represents a significant foray by Amazon into video generation.

According to AWS Developer Advocate Elizabeth Fuentes in a blog post, users can generate long videos composed of six-second shots by providing prompts up to 4,000 characters long. The new Nova Reel also introduces a "Multishot Manual" mode. In this mode, users can provide image and text prompts to help the model better control the composition of video shots. Fuentes notes that with a 1280x720 resolution image and a prompt of up to 512 characters, Multishot Manual mode can generate videos with up to 20 shots.

Nova Reel is currently only available through the Amazon AWS platform and related services (such as the AI development kit Bedrock), requiring customers to apply for special access. With more companies entering the video generation space, Nova Reel faces stiff competition from products offered by OpenAI, Google, and others.

However, questions remain about the model's training methodology. Generative AI models typically require training on vast datasets of video samples to "learn" to generate new video clips. Concerns exist that some companies may have used copyrighted videos for training without the permission of owners or creators, potentially exposing users to intellectual property lawsuits. While Amazon hasn't disclosed the source of Reel's training data or clarified how creators can opt out of its dataset, the company states it will protect any AWS customers accused of infringing on the copyrights of media generated by its model under its indemnification policy.

Official blog: https://aws.amazon.com/cn/blogs/aws/amazon-nova-reel-1-1-featuring-up-to-2-minutes-multi-shot-videos/

Key Highlights:
🎥 Nova Reel 1.1 can generate videos up to two minutes long and supports multi-shot production.
📊 Users can provide prompts up to 4,000 characters to generate videos with six-second shots, up to a maximum of 20 shots.
⚖️ Amazon's secrecy regarding the source of its training data has sparked discussion about copyright and intellectual property rights.

Amazon CEO Reveals Custom Chips Lowering AI Costs, $100 Billion Investment Planned for 2025

In a recent annual letter to shareholders, Amazon CEO Andy Jassy highlighted the company's significant investment in artificial intelligence (AI). He noted that while the development and deployment costs of AI remain high, future AI usage costs are expected to decrease significantly as technology advances. Image Note: Image generated by AI, image licensing provider Midjourney. Jassy revealed that Amazon plans to invest up to $100 billion in capital expenditures in 2025.

Reply and AWS Partner to Drive Generative AI Innovation

Reply, a leading global systems integrator and consulting firm, recently announced a multi-year strategic collaboration with Amazon Web Services (AWS) to accelerate innovation and adoption of generative AI (GenAI). This collaboration will empower global enterprises to leverage the potential of AI, utilizing advanced cloud infrastructure and AI-powered capabilities. Reply, along with its subsidiaries Data Reply and Storm Reply, will work with AWS to develop industry solutions that improve process efficiency and productivity.

Amazon Launches Nova Act: A New Push into the AI Agent Market

Amazon on Monday unveiled Nova Act, a general-purpose AI agent capable of controlling web browsers and independently performing simple operations. Also launched was the Nova Act SDK, allowing developers to build agent prototypes using Nova Act. Developed by Amazon's newly established AGI lab in San Francisco, Nova Act will also provide key functionality for the company's upcoming Alexa+ (the generative AI-enhanced version of Amazon's voice assistant). However, the currently released version of Nova Act is...

Ant Group Unveils EnergyTS, a Time-Series Model for Energy and Power, Surpassing Google and Amazon in Accuracy

On March 26th, at the Spring Summit of the New Energy Digital Asset Community in Suzhou, Ant Group officially launched EnergyTS, a time-series model for energy and power. This marks a significant breakthrough for Chinese companies in the field of energy AI. Specifically designed for the new energy sector, EnergyTS demonstrated significantly higher accuracy in photovoltaic power generation prediction compared to leading international general-purpose time-series models such as Google's (TimesFM-V2.0) and Amazon's (Chronos-Large). In T+1 day prediction, the model's mean absolute error is only...