Amazon Nova Reel hits 2-minute video generation with new multi-shot mode

Most AI video generators can give you a jaw-dropping five-second clip, but asking them to make a coherent two-minute video with matching styles usually ends in disaster.

Amazon wants to change that. AWS developer advocate Elizabeth Fuentes just revealed the rollout of Nova Reel version 1.1, and the headline number is a big one. The platform now supports AI video generation up to two minutes long, effectively doubling its previous one-minute limit.

Directing multi-shot scenes with one prompt

The real upgrade here is not just extra seconds. It is the new Multi-shot capability designed to maintain visual continuity across scene cuts.

Users can now input prompts up to 4,000 characters. Nova Reel then pieces together a series of six-second shots into a unified video that lasts up to two minutes, keeping the stylistic aesthetic consistent throughout.

For creators who want tighter reins, Amazon also introduced a Multishot Manual mode. You can feed it a single 1280x720 reference image alongside a 512-character prompt, and the model can produce a sequence featuring up to 20 shots based on that visual anchor.

Amazon’s AI balancing act

Amazon has been steadily pushing generative AI into practically every corner of its business, from the Rufus shopping assistant on its retail store to tools across Kindle and streaming platforms.

Yet its track record with recent consumer rollouts has been mixed. Earlier this year, Amazon launched Alexa+ with promises of advanced smart home routines and document analysis, but users found it underwhelming due to missing features previewed in earlier demos.

With Nova Reel 1.1, the cloud giant seems eager to prove that its media generation tech can deliver practical utility for enterprise and creative workflows.

Why it matters

We think Amazon made a very clever trade-off here. Generating a continuous two-minute single shot with AI is a nightmare for compute and consistency. Splitting the timeline into six-second modular shots stitched together from a single prompt or anchor image is a much more practical engineering answer for marketing teams and video creators.

That said, whether you should jump in right now depends on how much manual control you expect. Stitching 20 shots automatically is great on paper, but the real test is whether the transitions feel cinematic or like a slideshow of loosely related B-roll.

Still, by pushing the bar to multi-shot storytelling rather than brief single-shot visual flexing, Amazon is putting pressure on rivals like Runway, Luma, and OpenAI to make multi-scene consistency a default feature rather than a painful multi-step workflow.