Veo 3.1 update brings granular editing to Google's Flow tool

Google DeepMind released Veo 3.1, an updated version of its generative video model. This update focuses on offering greater creative control and richer native audio within the Flow filmmaking tool. Veo 3.1 and its new capabilities are immediately relevant for developers, becoming available via the Gemini API and Vertex AI.

The underlying Veo model, which powers Flow, has generated over 275 million videos since its introduction five months ago. The Veo 3.1 update addresses user feedback for better artistic control and complete audio support across all generation features.

Technical upgrades in Veo 3.1

Veo 3.1 improves on the Veo 3 model with stronger adherence to prompts and higher audiovisual quality when creating videos from images. It also delivers richer native audio, including synchronised sound effects and dialogue, giving videos a more polished feel.

The model is accessible to developers through the Gemini API and to enterprise customers through Vertex AI. This availability allows businesses to embed high-quality video generation directly into their applications and workflows. Pricing for Veo 3.1 Standard and Veo 3.1 Fast remains consistent with the Veo 3 model, costing developers USD 0.40 and USD 0.15 per second of generation, respectively.

Flow tool gains precision editing

Google's Flow, the dedicated AI filmmaking tool, receives significant updates powered by Veo 3.1. These updates move beyond simple video generation, adding tools for more granular scene development and editing.

Audio Across Existing Capabilities: The rich, generated audio now supports three established features, making them more powerful:

• • Ingredients for Video: Users provide multiple reference images to guide the style, characters, and objects in a scene. The generated video now includes contextual audio.

  • Frames to Video: Users define a precise starting and ending image. Flow generates the smooth transition video between them, now complete with audio. This is suitable for artful transitions.

  • Extend: This feature creates longer, continuous shots—lasting a minute or more—by generating a new clip that continues the action and background audio from the last second of the previous clip.

New Editing Tools for Clips: Flow introduces editing functions directly within the tool, addressing the need for mid-process adjustments:

• • Insert: Users add new elements, like creatures or detailed objects, to any scene. Flow adjusts scene lighting and shadows to make the additions appear natural within the clip.

  • Object Removal (Coming Soon): Flow will soon allow users to remove unwanted characters or objects. The system automatically reconstructs the scene's background to ensure a clean edit.

The addition of object insertion and removal capabilities, specifically designed for handling complex details like lighting and shadows, positions Flow as a more practical post-production tool. This level of control appeals directly to professional workflows where shot-by-shot perfection is essential.

Developer Access

The immediate availability of Veo 3.1 and the core creative controls, Ingredients to Video, Frames to Video, and Extend, through the Gemini API and Vertex AI is key for the B2B sector.

Companies building media applications, marketing platforms, or internal content pipelines can leverage these updated models for:

1. Consistent Asset Generation: Using Ingredients to Video with reference images allows businesses to maintain character and product consistency across a series of generated marketing clips or training materials.

2. Automated Storyboarding and Previsualization: Tools like Frames to Video and Extend enable media studios to rapidly prototype complex camera moves and longer narrative sequences, speeding up pre-production cycles.

3. Scalable Content Production: By offering Veo 3.1 through Google's enterprise platform, Vertex AI, Google makes the technology available for high-volume production needs, maintaining pricing parity with the previous version.

The focus on refined generation, complete with native audio, signals Google's intent to move Veo from an experimental technology toward a tool for production-ready content creation.