Veo 3.1 is Google DeepMind's current flagship AI video model, and it has one capability that no other model at this level matches: genuinely integrated audio. Not audio layered over silent footage, but ambient sound, dialogue, and environmental audio generated in sync with the visual content as part of the same output. For marketers making ad creative in 2026, that distinction matters more than it might initially seem. This guide covers what Veo 3.1 is good for, how to prompt it effectively, how it compares to Kling 3.0, and how to access it without a $250/month Google subscription.
What makes Veo 3.1 different from other AI video models
Audio is the defining capability
Every serious AI video model generates visually plausible footage. The differentiator for Veo 3.1 is audio synchronisation. When a product is placed on a surface in your video, you hear the contact sound. When wind appears in the scene, you hear it at the right intensity. When a character speaks, the voice is timed to the visible movement. This is not background music or generic ambient noise — it is scene-specific audio that corresponds to what is actually happening in the footage.
This matters for ad creative because social platforms — TikTok, Instagram Reels, YouTube Shorts — are audio-on environments by default. Viewers notice when audio feels detached from visuals. Veo 3.1 eliminates that friction. You get a complete asset: visual and audio generated together, ready to post without a separate sound design pass.
Prompt precision
Veo 3.1 has unusually high prompt adherence. Describing a specific visual — a matte black product against a linen-textured background, lit from a single window — produces output that reliably reflects those instructions. For brand creative where you have defined visual standards, this predictability is practically useful. You can build and refine a prompt framework and get consistent results rather than running dozens of generations hoping the model interprets the brief correctly.
Photorealism and material rendering
Veo 3.1 handles materials and lighting at a high level. Glass, metal, fabric, skin — the surface rendering is accurate and physically convincing. For product-forward ad creative, this is important: you need textures to look real, not like a 3D render approximation. Veo 3.1 handles this well across a wide range of surface types and lighting setups.
What Veo 3.1 is best for
- Social video with native audio — TikTok and Reels content where removing audio is not an option. Veo 3.1 generates both layers as one coherent asset.
- DTC product videos — Lifestyle shots of products in use, with the ambient sounds of the setting (kitchen sounds, outdoor environment, café background) giving footage authenticity.
- Lifestyle brand content — Environmental scenes — morning routines, outdoor settings, workspace environments — where the combination of visual and audio atmosphere sells a feeling.
- Ad creative where audio is part of the message — ASMR-style content, product interaction moments, any ad where sound design is deliberate rather than decorative.
- Structured prompt workflows — Teams with defined prompt libraries and creative briefs that need high fidelity output, not creative interpretation.
How to prompt Veo 3.1 effectively
Veo 3.1 responds well to structured prompts with five components: subject, scene, audio, camera, and mood. Treating audio as an explicit prompt element — rather than hoping the model infers it — produces significantly better results.
Prompt structure
- Subject — What is in the frame. Be specific: product name, material, colour, size relative to frame.
- Scene — The environment. Surface, background, props, time of day, interior or exterior.
- Audio — Describe the sounds explicitly. What should the viewer hear? Ambient setting, product interaction sounds, background noise level.
- Camera — Movement, angle, and distance. Static or moving. Wide, medium, or close. Slow push-in, orbit, handheld feel.
- Mood — The overall register. Aspirational, minimal, warm, clinical, energetic.
Example prompts for ad creative
DTC skincare product — social video:
"A glass serum bottle placed on a white marble surface. Morning bathroom setting with soft diffused light through frosted glass. Sound of the bottle being set down on marble, quiet ambient room tone, distant birds outside. Slow push-in from medium to close, 8 seconds. Mood: clean, premium, calm."
Lifestyle brand — coffee morning content:
"Person wrapping hands around a ceramic mug, steam rising, sitting at a wooden kitchen table. Morning light coming in from the left, warm colour temperature. Sound of coffee being poured, spoon briefly stirring, quiet background hum of morning. Handheld feel, slightly warm lens. Mood: comfortable, unhurried, aspirational."
Activewear — outdoor lifestyle:
"A runner moving along a coastal path, cliffs visible in background. Early morning golden hour. Sound of footsteps on gravel path, rhythmic breathing, distant wind and waves. Tracking shot from the side, keeping pace. Mood: focused, free, energetic."
Notice that in each example, audio is explicitly described rather than left to the model. Prompts that treat audio as an afterthought — or omit it entirely — tend to produce generic ambient sound rather than scene-specific audio that adds to the creative.
Veo 3.1 vs Kling 3.0: when to use which
These are currently the two strongest AI video models available. They are not equivalent — they each have areas where they outperform the other, and choosing the right one for a brief matters.
Use Veo 3.1 when:
- Audio is a meaningful part of the creative brief — product sounds, environmental atmosphere, voice
- You have a detailed, structured prompt and want high fidelity adherence to specific visual instructions
- The content is product-focused, lifestyle-orientated, or environmental (as opposed to narrative or action-heavy)
- You are making content for audio-on platforms and want a complete asset without post-production audio work
Use Kling 3.0 when:
- You need complex narrative scenes — multiple subjects, story progression across a clip
- Photorealistic cinematic quality is the primary requirement and audio is secondary
- The content involves physical action — sports, product motion, fast-moving subjects
- You want motion control or image-to-video generation with reference image input
In practice, most production workflows use both. Veo 3.1 for the product and lifestyle close-ups where audio matters; Kling 3.0 for the wide narrative and action sequences. Having both on the same platform makes this straightforward rather than requiring separate accounts.
Use case examples
DTC brand: skincare launch
A direct-to-consumer skincare brand launching a new serum needs five ad variants for Meta and TikTok: product close-ups, lifestyle application shots, morning routine context. Veo 3.1 is the right model here. Each variant can include the ambient sounds of the environment — bathroom acoustics, product sounds, morning texture — giving the ads a production quality that stock video rarely achieves without a full production team. Prompt frameworks can be built around the brand's visual standards and reused across the campaign.
Lifestyle brand: seasonal content
A clothing brand making seasonal content for Instagram Reels needs footage that communicates atmosphere more than product detail. Outdoor settings, morning light, textured environments. The audio layer — wind, fabric movement, ambient natural sound — is what makes this content feel authentic rather than generated. Veo 3.1's ability to generate that audio layer as part of the output reduces post-production time and makes the content feel cohesive rather than assembled.
Social media ads: food and beverage
Food and beverage brands have always relied on sound as a selling tool — the sizzle, the pour, the crunch. Veo 3.1 can generate this. A coffee brand can produce a 10-second video of a pour shot with the exact audio of a high-quality espresso being extracted, all generated as one asset. For a category where audio-visual appeal is the mechanism of purchase intent, this is a significant capability.
How to access Veo 3.1 without a Google subscription
Google's own access to Veo 3.1 is available through their Flow product at approximately $250 per month — a subscription designed for enterprise users and professional video producers. For brands and marketers who want Veo 3.1 as one tool among several rather than as a full platform investment, this pricing structure is impractical.
Xarith provides access to Veo 3.1 on a credit-based model. You pay per generation rather than a flat monthly fee, and you get Veo 3.1 alongside Kling 3.0, the full Kling family, and 14+ other frontier video and image models on a single account. For production teams that need to switch between models based on the brief — and do not want to maintain separate subscriptions for each — this is the practical access route.
See the pricing page for current credit costs per model. Veo 3.1 Fast is also available for faster, lower-cost iterations during the prompt development phase — the workflow recommendation is to iterate with Veo 3.1 Fast until the prompt is dialled in, then generate finals with standard Veo 3.1.
Key things to know before you start
- Describe audio explicitly — do not leave it to inference. Treat sound as a first-class element of your prompt, not an afterthought.
- Be specific about materials and lighting — Veo 3.1 has the precision to execute detailed visual instructions. Vague prompts produce vague results; specific prompts produce specific results.
- Text in video remains unreliable — do not attempt to generate readable on-screen text across any current AI video model, including Veo 3.1.
- Single subjects outperform groups — complex scenes with multiple interacting characters are harder for all current models. Single-subject and environmental content performs most reliably.
- Iterate on Fast, finalise on Standard — use Veo 3.1 Fast to develop your prompt, then move to standard Veo 3.1 for final outputs to manage both speed and cost.
Veo 3.1 is not the right model for every brief. But for audio-forward social content, DTC product video, and lifestyle brand creative where the sound design is part of the asset — it is the best AI video model available as of March 2026. The prompting approach is learnable quickly, and the output quality justifies the time investment in building a prompt framework around it.