Veo 3.1 Lite Just Dropped Video Generation to $0.05 a Second. Here Are Three SaaS Products You Could Ship This Month That Weren't Viable Last Year.
At Google I/O 2026 on May 19, Google restructured Veo's pricing into three tiers: Lite, Fast, and Standard. Veo 3.1 Lite starts at $0.05–$0.08 per second of generated video. The previous pricing structure didn't have a Lite tier: it was Fast and Standard, with Fast running around $0.25–$0.40/second depending on resolution.
That's a 5x price reduction for the entry tier. And it's accessible through the Gemini API today, billable by the second, with no sales call or enterprise contract required.
A 60-second product demo video: $3–$5. A 3-minute explainer: $9–$15. A 10-second social clip: $0.50–$0.80.
The question isn't whether the math is interesting. It's whether the math changes which products a solo builder can actually ship.
What changed and what didn't
The important distinction with Veo 3.1 Lite is what it's good at and where it falls short, because the tradeoffs matter for product decisions.
Veo 3.1 Lite produces video that is good enough for: product demos, explainer content, social shorts, automated visualization of data or events, personalized onboarding clips. It is not good enough for: cinematic production quality, photorealistic human faces at length, precise motion control where exact behavior matters.
The "not good enough for" list is shrinking fast. Six months ago, Veo's entry tier was not usable for anything customer-facing. At Lite tier pricing and quality, there are specific product categories where the output is above the bar.
The other thing that changed: Google moved Veo into the standard Gemini API infrastructure. This means you can mix text generation, audio, and video generation in the same API call chain. Build an agent that writes a script, generates audio narration, and produces a video: one API key, one billing relationship, one set of rate limits to manage. Six months ago you were integrating ElevenLabs for audio, Runway or Pika for video, and an LLM for the script. Now it's one place.
Three products that cross the viability line
I want to be concrete about product shapes rather than abstract about "use cases."
Automated product demo videos for SaaS signups. The pattern: new user signs up, your backend generates a 60-90 second personalized demo video showing their specific use case or industry, sends it as part of the onboarding sequence. The video isn't generic: it references the user's company name, their industry vertical, the features most relevant to their signup data. You couldn't build this affordably when video generation cost $0.40/second because a 90-second video cost $36 per user. At $0.07/second, that's $6.30 per user. If your product has any meaningful LTV, that math works. You build the script generator, call Veo 3.1 Lite, store the video in cloud storage, send it via email or in-app. One weekend to ship a basic version.
Personalized social video for e-commerce product pages. Show me a 10-second clip of the product in a context that matches my apparent interests. Standard product page video is expensive to produce at scale across a catalog. AI-generated video at $0.07/second makes per-product, even per-variant video economical. A 500-product catalog at 10 seconds each: $35 in video generation costs. Refresh it quarterly: $140/year. This is a feature e-commerce SaaS tools could charge for that previously required bespoke video production.
Automated social clips from blog posts or newsletters. The pattern: take a piece of written content, generate a 30-60 second vertical video that adapts the key points into a social-native format with generated visuals, captions, and a hook in the first 3 seconds. Tools like this exist already (Pictory, Opus Clip) but they rely on stock footage or require uploaded source video. A pure text-to-video version that generates novel visuals is a different product. At $0.07/second, a 45-second clip costs $3.15 in video generation. If you charge $49/month for 15 clips, your video generation cost is $47 and your margin is $2. That's not a great business. If you charge $149/month for 50 clips, your video generation cost is $157 and you're underwater. You'd need to run the numbers carefully on compression (Lite tier's quality can be pushed further with good prompts), caching (if two users want similar content, serve the same video), and batching. This one is viable but tighter than it looks.
Where the math still doesn't work
There are product categories where $0.07/second is still too expensive.
Long-form video at scale. If your product generates 10-minute videos for users, each video costs $42 at Lite tier. At 1,000 users generating one video per month, that's $42,000/month in video generation costs. You'd need to charge $100+ per user just to break even on video. This is a different market than the mass-market SaaS I'm describing.
Real-time video generation. Veo 3.1 Lite is not a real-time API. Generation latency is measured in seconds to minutes depending on length and complexity. If your product needs to generate video as part of a synchronous user interaction, the current generation speed doesn't support that product shape. You'd need to pre-generate and cache, or find a use case where the user initiates the generation and comes back for the result.
Precision motion. If you need a video where specific things happen at specific times (a character walks to a specific spot, picks up a specific object, performs a specific action), Lite tier (and honestly any current tier of Veo) isn't the right tool. The model interprets prompts directionally, not precisely. For explainer and demo content this is fine. For anything requiring precise action sequencing, you're still in the territory of recorded video.
The API call pattern
For builders who want to see the actual implementation shape, here's the rough structure for the automated onboarding demo use case:
# After user signs up, in your onboarding webhook
script = generate_script_with_gemini(
user_industry=user.industry,
user_role=user.job_title,
product_features=relevant_features_for_user(user)
)
video_response = veo_client.generate_video(
prompt=script,
duration_seconds=90,
aspect_ratio="16:9",
model="veo-3.1-lite"
)
video_url = upload_to_storage(video_response.video_bytes)
send_onboarding_email(user.email, video_url=video_url)
Actual implementation is more verbose: you'll want async handling since generation takes time, retry logic, and a job queue for anything at scale. But the surface area is small. This is a weekend project, not a sprint.
The honest take
The Lite tier pricing is genuinely interesting for the three product shapes I described. It's not transformative for everything: long-form video economics are still hard, precision motion is still hard, real-time is still hard. But the specific window that opened is: automated short-form personalized video as a feature embedded in an existing SaaS product, where video generation is one step in a larger workflow rather than the core product.
The companies who will overpay for this mistake are the ones who see the $0.05/second headline and build their entire product around video generation without modeling the actual unit economics at their target scale. Do the math before you build. But if the math works at your scale and your price point, the product is now shippable.
Author
Lukas
@lukcombinator