Amazon has quietly taken a big step forward in the AI video race. The tech giant has updated its generative video model, Nova Reel, to create clips lasting up to two minutes—a move that could intensify its competition with rivals like OpenAI and Google.
First launched in December 2024, Nova Reel marked Amazon’s initial entry into the fast-growing generative video space. Now, with the latest upgrade—Nova Reel 1.1—the tool offers a major performance boost and new features aimed at giving users more creative control.
Nova Reel 1.1 Brings Longer, More Cohesive AI Videos
According to a blog post from AWS developer advocate Elizabeth Fuentes, Nova Reel 1.1 is capable of generating “multi-shot” videos that maintain a consistent visual style across scenes. The tool now accepts prompts of up to 4,000 characters to produce two-minute-long videos made up of six-second shots. This means creators can craft more detailed narratives and visual sequences without abrupt style shifts.
New Manual Mode Offers Greater Creative Control
A standout feature of this update is the introduction of “Multishot Manual”—a new mode that lets users pair images with prompts to shape the video’s composition more precisely. By combining a 1280 x 720-pixel image and a short text prompt of up to 512 characters, the system can generate videos with up to 20 unique shots.
This mode aims to bridge the gap between creative intent and machine output, giving users a semi-guided way to direct how scenes should look and feel. It’s an important feature for designers, marketers, and content creators seeking visual consistency without starting from scratch each time.
Available Only Through AWS—With Some Restrictions
Nova Reel remains exclusive to Amazon’s AI infrastructure. Access is currently limited to AWS platforms and tools, including Amazon Bedrock—its suite for building and deploying generative AI. Users must request access to the model, suggesting Amazon is still keeping the system somewhat gated, possibly to manage responsible use or scale.
Data Transparency Questions Remain
As with many generative AI systems, questions persist around how Amazon trained Nova Reel. Creating AI video models typically involves feeding systems with massive datasets of video content. But if that content includes copyrighted materials obtained without consent, it opens the door to legal and ethical issues.
Amazon has not disclosed the specific sources used to train Nova Reel, nor has it provided a way for creators to opt out of having their content used. This lack of transparency puts the company in line with several competitors but may concern creators whose videos could have been used to train the model.
Still, Amazon has said it will indemnify AWS customers using Nova Reel from potential copyright claims. This promise echoes similar commitments from other cloud giants aiming to encourage business adoption while minimizing legal risk.
