As expected after days of leaks and rumors on the web, Google did just that presented I Spy 3.1the latest AI video generation model, offering a suite of creative and technical enhancements aimed at improving narrative control, audio integration and realism of AI-generated video.
While the updates expand the capabilities of hobbyists and content creators using Google’s online AI app, Flowthis release also signals growing opportunities for enterprises, developers and creative teams looking for scalable and customizable video tools.
The quality is higher, the physics are higher, the prices are the same as before, and the control and editing features are more robust and varied.
My preliminary tests has shown that it is a powerful and efficient model that immediately delights every generation. However, the look is more cinematic, polished and a bit more “art” than the default than rivals like OpenAI’s recent Sora 2, released late last month, which can or may not suit a particular user (Sora excels in mobile video and “honest” style).
Expanded control over narration and audio
Veo 3.1 is based on its predecessor, Veo 3 (released May 2025) with improved support for dialogue, ambient sounds, and other sound effects.
Native audio generation is now available in several key Flow features, including “Frames to Video,” “Components to Video,” and “Augmentation,” which give users the ability to: transform still images into video; use objects, characters and objects from multiple images in one video; and generate clips longer than the initial 8 seconds, as much as over 30 seconds, or even 1+ plus when continuing from the last frame of the previous clip.
Previously, you had so as to add audio manually after using these features.
This addition gives users greater control over tone, emotion and storytelling – capabilities that previously required post-production work.
In a corporate context, this level of control can reduce the need for separate audio pipelines by offering an integrated option to create training content, marketing videos, or digital experiences with synchronized audio and video.
Google noticed in blog post that the updates reflect user feedback calling for greater artistic control and higher audio handling. Gallegos emphasizes the importance of having the ability to edit and refine directly in Flow, without having to redo scenes from scratch.
Richer input and editing capabilities
In Veo 3.1, Google introduces support for multiple input types and more granular control over the output generated. The model accepts text prompts, images, and video clips as input, and supports:
-
Reference images (maximum three) to guide the look and feel of the final product
-
Interpolation of the first and last frame to generate seamless scenes between fixed endpoints
-
Scene expansion that continues the motion or movement of a film beyond its current running time
These tools are designed to present enterprise users the ability to customize the look and feel of their content – useful for ensuring brand consistency or adhering to creative briefs.
Additional capabilities comparable to “Insert” (adding objects to scenes) and “Delete” (removing items or characters) are also being introduced, although not all are immediately available via the Gemini API.
Implementation on various platforms
Veo 3.1 could be accessed through several existing Google AI services:
-
FlowGoogle’s own AI-powered video creation interface
-
Gemini APIaimed at developers building video capabilities in applications
-
Apex AIwhere enterprise integration will soon support Veo “Stage Extension” and other key features
Availability on these platforms allows enterprise customers to decide on the right environment – GUI-based or software-based – based on their teams and workflows.
Prices and access
Currently, the offer includes the Veo 3.1 model announcement and available exclusively on paid tier Gemini API. The cost structure is the same as Veo 3, Google’s previous generation of AI video models.
-
Standard model: $0.40 per second of video
-
Fast model: $0.15 per second
There is no free tier and users are only charged if the video is successfully generated. This model is consistent with previous versions of Veo and provides predictable pricing for budget-conscious enterprise teams.
Technical data and performance control
Veo 3.1 sends video in 720p or 1080p resolutionWith Frame rate 24 fps.
Duration options include 4, 6 or 8 seconds from text prompts or uploaded images, with the ability to expand videos to 148 seconds (over 2 and a half minutes!) when using the “Extend” function.
The recent functionality also includes tighter control over objects and the environment. For example, businesses can upload a product photo or visual reference and Veo 3.1 will generate scenes that maintain its look and style cues throughout the video. This can streamline creative production processes for retail, promoting and virtual content production teams.
Initial reactions
The broader creative and developer community responded to the launch of Veo 3.1 with a mixture of optimism and moderate criticism – especially when in comparison with competing models comparable to Sora 2 from OpenAI.
Matt Shumer AI founder at Otherside AI/Hyperwrite and an early adopter, described his initial response as “disappointment,” noting that Veo 3.1 is “noticeably inferior to Sora 2” in addition to “quite a bit more expensive.”
However, he admitted that Google tools – comparable to reference handling and scene expansion – are the uniqueness of this version.
Travis Davids3D digital artist and AI content creator, shares some of those opinions. While he noted improvements in audio quality, particularly in sound effects and dialogue, he expressed concerns about limitations still present in the system.
These include the lack of custom voice support, the inability to directly select generated voices, and the continued limitation to 8-second generations – despite some public claims of longer results.
Davids also identified that character consistency across changing camera angles still requires careful prompting, while other models like Sora 2 handle this more mechanically. He questioned the lack of 1080p resolution for users of paid tiers like Flow Pro and expressed skepticism about feature parity.
From a more positive standpoint, @kimmonismus, the writer of the AI newsletter stated that “Veo 3.1 is amazing”, although he nevertheless stated that the latest OpenAI model is generally preferred.
Collectively, these first impressions suggest that while Veo 3.1 delivers significant tool improvements and recent creative control features, expectations have modified as competitors raise the bar in each quality and usability.
Adoption and scale
Since launching Flow five months ago, Google has called it quits 275 million videos have been generated for various Veo models.
The pace of implementation indicates great interest not only from private individuals, but also from developers and corporations experimenting with automatic content creation.
Thomas Iljic, director of product management at Google Labs, emphasizes that the release of Veo 3.1 brings capabilities closer to how filmmakers plan and shoot. These include scene composition, shot continuity and coordinated audio – all areas that enterprises are increasingly seeking to automate or streamline.
Safety and responsible use of artificial intelligence
Videos generated with Veo 3.1 are watermarked by Google SynthID technology that features a transparent identifier that indicates that the content has been generated by artificial intelligence.
Google uses security filters and moderation in its APIs to attenuate privacy and copyright risks. Generated content is temporarily stored and deleted after two days unless downloaded.
For developers and enterprises, these features provide provenance and compliance assurance, which is crucial in regulated or brand-sensitive industries.
Veo 3.1 stands out in a crowded AI video model space
Veo 3.1 is not only an iteration of previous models – it represents deeper integration of multimodal inputs, storytelling control and enterprise-level tools. While creative professionals may see immediate advantages in workflow editing and fidelity, corporations in search of automation in training, promoting, or virtual experiences may find even greater value in model composability and API support.
Early user feedback shows that while Veo 3.1 offers worthwhile tools, expectations regarding realism, voice control and generation length are rapidly evolving. As Google expands access through Vertex AI and continues to refine Veo, its competitive position in enterprise video generation will depend on how quickly user problems are resolved.