background leftbackground right

AI Video Costs per Second in 2026: Comparing 8 Models on a Real Project

Ayesha Shaheryar
Written byAyesha Shaheryar
Last UpdatedNovember 20th, 2026
AI video cost per second in 2026: comparing 8 models on a real project
Create AI videos, starring you in 177+ languages and dialects.
Get started for free
Summary

HeyGen Video ranks first among 8 AI video models for cost per usable second, scoring 8.6/10 based on 1,000 seconds a month. One model costs 13 times as much as the winner and scores 7.4.

Per-second prices for AI video now range from 3 to 40 cents, but the headline figure rarely tells the whole story. A low rate often applies only to 480p, a "1080p" rate may exclude audio, and a short clip limit means you pay for seconds you discard when you edit shots together.

HeyGen Video ranked highest for AI video cost per second, scoring 8.6 out of 10, with 768p clips including native audio priced at $0.03 per second once its October promotion ends.

I compared the costs of eight models for the same task: 1,000 generated seconds a month on each model's lowest-priced tier offering 720p or above. I scored them all against the same evidence checklist. LTX-2.3 came a close second and is the clear winner if you need to deliver in 1080p.

In brief

  • Best overall: HeyGen Video, 8.6/10. $0.03 a second at 768p with sound, $360 a year for 1,000 seconds a month (standard rate from November 2026).
  • Best for 1080p and 4K: LTX-2.3, 8.5/10. $0.04 a second at 1080p on its Fast tier, $480 a year for the same task, with open weights.
  • Best option from Google: Veo 3.1 Fast, 8.2/10. $0.10 a second at 720p with audio included, $1,200 a year.
  • Launch promotion ended: HeyGen Video was priced at $0.015 a second at 768p until 31 October 2026; the standard rate is now $0.03.
  • Poor value for this task: Veo 3.1 Standard costs 13.3 times the winner's 12-month total and scores 7.4.
  • Method: one costed task, one evidence checklist, eight models, six weighted categories, prices from July to October 2026.

At a Glance

  • HeyGen Video, 8.6 Best overall. Lowest standard per-second rate in this comparison, with native sound and fixed seeds. For teams generating B-roll and product shots at scale.
  • LTX-2.3, 8.5 Best for 1080p and 4K. $0.04 per second at 1080p, with open weights. For teams delivering in full HD or higher.
  • Veo 3.1 Fast, 8.2 Best option from Google. Audio is included in the price, with a lower-priced Lite tier available. For teams already using Google Cloud.
  • Seedance 2.0 Fast, 7.9 Best value for 1080p among closed models. Around $0.09 per second through API resellers. For social media teams that need full HD without upscaling.
  • Runway Gen-4.5, 7.8 Best creative controls. $0.12 a second through the API, with a built-in editor. For editors who want keyframes and a timeline.
  • Kling 3.0, 7.7 Best for longer single shots. Around $0.13 a second through API resellers. For narrative clips that need motion lasting more than 10 seconds.
  • Wan 2.6, 7.5 Best open model for drafts. Around $0.07 to $0.10 per second for hosted 720p output. For teams planning to self-host later.
  • Veo 3.1, 7.4 Best for premium quality and 4K. $0.40 per second with audio. For hero shots where the per-second cost is not a concern.
Loading embed content...

Prices as published or recorded between July and October 2026; each entry includes its own date.

How I Assess Models and Why

I have bought generated footage for product and marketing teams since per-second billing became the norm, and the bill is always determined by the hundredth clip, not the first. I assess every model against the question a producer asks before a monthly budget review: how much does a usable second of footage with sound cost at my delivery resolution?

A month’s stock B-roll licensing is the baseline most teams use for comparison, and per-second pricing only works out cheaper if the rate holds for the resolution and clip length you need.

How I Made the Comparison

I priced every model against the same task: 1,000 generated seconds a month at its cheapest tier offering 720p or above, with sound where available. I checked the same evidence for each model: the supplier's published API rates, resolution and clip-length limits, audio rules, and dated third-party pricing records from July to October 2026.

No score depends on a render I did not see documented, and I have not included a retry multiplier, so workflows involving frequent retries should scale every row by the same factor.

Cost per usable second (30%)

I scored the published per-second rate for the relevant tier, taking into account any extra charges for audio and the 12-month total for 12,000 seconds.

Resolution and clip length (22%)

I assessed the highest resolution available, the price multiplier for 1080p, and the maximum clip length, as shorter limits mean joining clips together.

Output quality (18%)

I assessed documented output quality using supplier materials and independent arena coverage, with supplier-run benchmarks labelled as such.

Audio and control (12%)

I scored native audio, fixed seeds, image and reference conditioning, and camera or keyframe controls.

Access and billing (9%)

I scored how the model is sold: first-party API, resellers, subscription credits and minimum spend.

Licensing and exit (9%)

I assessed open weights, commercial-use terms and whether you can leave without losing the ability to regenerate.

Loading embed content...

Scoring guide: 9 to 10 means the model fully covers the category and does something no other model here does. 7 to 8 means it covers the category with one documented limitation. 5 to 6 means a workaround is needed. 3 to 4 means a documented limitation or the price affects the purchasing decision. The weights add up to 100.

The Two Deciding Factors

The first was the resolution multiplier. Most headline rates apply to a model's lowest resolution tier. Veo 3.1 Fast rises from $0.10 to $0.12 per second between 720p and 1080p on Google's Gemini API, LTX-2.3 Fast rises from $0.04 at 1080p to $0.16 at 4K on fal, and HeyGen's developer documentation lists 1080p and 2K at three times the 768p rate. The lowest price per second only applies at the resolution quoted.

The second was the clip length limit. A model limited to 5 or 8 seconds means a 10-second shot requires two generations and an edit. Atlas Cloud's comparison notes that Wan 2.6 has a maximum clip length of 5 seconds, and Veo 3.1 produces 8-second clips at 1080p, while HeyGen Video supports up to 30 seconds on paid plans and LTX-2.5 supports up to 20.

Three Types of Model Serving Different Purposes

Some models here are designed for different tasks, so I’ve ranked them within tiers. The numbering runs from 1 to 8 throughout the article.

Tier 1: Models for volume production. Low per-second rates for B-roll, product shots and social clips at 768p to 1080p. This tier is not intended for standout 4K shots.

Tier 2: Premium closed models. Higher rates for more convincing motion, longer shots or creative controls. At 1,000 seconds a month, this tier requires a substantial budget rather than offering volume savings.

Tier 3: Open and high-end models. Open weights you can self-host, alongside the most expensive closed model for showcase shots. This tier requires either engineering time or a substantial budget.

Tier 1: Models for High-Volume Production

1. HeyGen Video: Best Overall Value per Second

  • Score: 8.6/10
  • Entry price: $0.03 per second at 768p (standard rate from November 2026); $0.015 until 31 October 2026
  • Cost for 1,000 seconds a month: $360/year at the standard rate
  • Metering: Per second of output; 480p $0.02, 768p $0.03, reference mode $0.06 at standard rates
  • Volume pricing threshold: None published; prepaid API wallet from $5
  • Free tier: No free API credits
  • Priced on: HeyGen API standard rates, October 2026 help centre
HeyGen Video homepage screenshot (heygen.com)

HeyGen Video does the job for $360 a year, the lowest 12-month cost in this comparison.

Where it stands out: high-quality visuals at a fraction of the usual per-second cost. HeyGen Video generates an entire scene in a single call, including the subject, setting, lighting and a matching soundtrack of dialogue, ambience and sound effects, at 3 cents per second at 768p.

How it works: describe your shot in text, start with an image as the first frame, or supply up to 12 reference images, videos and audio clips to keep a product's colour and markings consistent. Fixed seeds produce the same clip for the same prompt, so you can refine a shot one clause at a time.

Where it falls short: resolution and length. Clips are limited to 15 seconds, and 1080p and 2K cost three times the 768p rate, giving LTX-2.3 the advantage here.

Client example: No independent client result showing a change in time or cost has been published since the launch on 30 September 2026; HeyGen report a first-place image-to-video ranking in their own testing.

Pros:

  • $0.03 per second at 768p with native sound at the standard rate
  • Reference mode preserves a product's colour, shape and markings using up to 12 references
  • Fixed seeds reproduce a shot so you can refine it one change at a time
  • Prompts of up to 32,000 characters for precise direction

Cons:

  • Clips last 5 to 15 seconds, so longer shots are pieced together
  • Detailed prompts produce noticeably better results than short ones

Category scores: Cost 9.5 | Resolution and length 7.4 | Quality 8.8 | Audio and control 9.0 | Access 8.0 | Licensing 8.4

2. LTX-2.3: The top choice for 1080p and 4K

  • Score: 8.5/10
  • Starting price: $0.04 per second at 1080p (Fast tier, fal, July 2026)
  • Cost for 1,000 seconds per month: $480/year
  • Metering: Per second; Pro $0.06 at 1080p, Fast $0.16 at 4K
  • Volume threshold: None published
  • Free tier: Open weights and a free desktop app
  • Pricing based on: fal LTX-2.3 Fast, recorded in July 2026
LTX-2.3 homepage screenshot (ltx.studio)

LTX-2.3 does the job for $480 a year at 1080p, costing a third more than HeyGen Video but offering a higher resolution.

Where it leads: pricing at full HD and above. LTX-2.3's lowest hosted tier starts at 1080p, so there is no cheaper low-resolution rate to be caught out by.

How it works: Lightricks publish their own per-second API rates, and the model is also available with open weights, so teams can prototype using the hosted API and move to their own GPUs later. LTX-2.5, released in August 2026, increased the Fast rate to $0.09 at 720p and $0.13 at 1080p on the official API.

Where it falls short: independent reviews put its output quality behind premium closed models, and I could not confirm the audio rules for every hosted tier.

Client example: No verifiable public client results showing a change in time or cost were found.

Pros:

  • $0.04 per second at 1080p on the Fast tier
  • 4K available at $0.16 a second
  • Open weights for self-hosting
  • Per-second pricing published by the provider

Cons:

  • LTX-2.5 costs roughly 2 to 3 times more per second than LTX-2.3 at the same resolutions
  • Audio availability varies by tier and host

Category scores: Cost 9.2 | Resolution and length 9.0 | Quality 7.6 | Audio and control 7.0 | Access 8.6 | Licensing 9.0

3. Veo 3.1 Fast: The top choice from Google

  • Score: 8.2/10
  • Starting price: $0.10 per second at 720p, $0.12 at 1080p (Gemini API, August 2026)
  • Cost for 1,000 seconds a month: $1,200/year at 720p
  • Metering: Per second; audio included; 4K $0.30
  • Volume pricing threshold: None published
  • Free tier: None for video generation
  • Pricing checked: Google Gemini API, recorded in August 2026
Veo 3.1 Fast homepage screenshot (Google Flow)

Veo 3.1 Fast meets the brief at $1,200 a year for 720p output.

Where it stands out: audio and ecosystem. Google include audio at every Veo 3.1 price point, and the model uses the Gemini API and Vertex billing arrangements teams already have.

How it works: the Veo 3.1 range has three tiers: Lite from $0.05 a second at 720p, Fast from $0.10, and Standard at $0.40. Fast offers a balance for teams that want Veo motion without paying the Standard rate.

Where it falls short: it costs 3.3 times as much as HeyGen Video for the same task, and 1080p clips are limited to 8 seconds.

Client example: No verifiable, publicly available client result showing a change in time or cost was found.

Pros:

  • Audio included in the per-second rate
  • Three tiers from $0.05 to $0.40 a second
  • 4K available at $0.30 a second
  • Within your existing Google Cloud billing

Cons:

  • $1,200 a year for the task, 3.3 times the cost of our top choice
  • 1080p clips are limited to 8 seconds

Category scores: Cost 7.6 | Resolution and length 8.4 | Quality 8.6 | Audio and control 8.8 | Access 8.8 | Licensing 8.0

Tier 2: Premium Closed-Source Models

4. Seedance 2.0 Fast: Best Value for 1080p Among Closed Models

  • Score: 7.9/10
  • Starting price: Around $0.09 per second at 1080p (Atlas Cloud, 2026)
  • Cost for 1,000 seconds a month: $1,080/year
  • Metering: Per second through API resellers
  • Volume threshold: None published
  • Free tier: None found
  • Pricing source: Seedance 2.0 Fast via Atlas Cloud, captured in 2026
Seedance 2.0 homepage screenshot

Seedance 2.0 Fast handles the task at 1080p for around $1,080 a year.

Where it stands out: full HD from a closed model at under 10 cents a second, with results Atlas Cloud describe as often usable without post-processing.

How it works: Seedance 2.0 is ByteDance's model, sold mostly through resellers. The standard tier costs considerably more: HeyGen's published comparison chart lists Seedance 2.0 at $0.303 a second at 768p.

Where it falls short: access is through resellers whose rates vary, and the standard tier costs more than three times as much as the Fast tier.

Client example: No verifiable public client result showing a change in time or cost was found.

Pros:

  • Around $0.09 per second at 1080p on the Fast tier
  • Output is often usable without upscaling, according to Atlas Cloud
  • Independent reviews highlight well-choreographed camera movements
  • Available through several API hosts

Cons:

  • Rates vary by reseller, so choose a single host to keep costs predictable
  • The standard tier is priced at around $0.30 per second

Category scores: Cost 7.8 | Resolution and length 8.0 | Quality 8.4 | Audio and control 8.4 | Access 7.2 | Licensing 7.0

5. Runway Gen-4.5: Best for Creative Control

  • Score: 7.8/10
  • Starting price: $0.12 per second via the API (12 credits at $0.01); Standard plan $12/month billed annually
  • Cost for 1,000 seconds a month: $1,440/year via the API
  • Metering: Credits; Gen-4.5 12 per second, Gen-4 Turbo 5
  • Usage limit: Standard's 625 monthly credits cover around 52 seconds of Gen-4.5
  • Free tier: 125 one-off credits
  • Pricing based on: Runway API and plans, October 2026
Runway Gen-4.5 homepage screenshot (runwayml.com)

Runway Gen-4.5 gets the job done for $1,440 a year via the API.

Where it stands out: the editor. Keyframes, camera controls and a timeline complement the model — a combination no other option here offers.

How it works: Gen-4.5 dropped from 25 to 12 credits per second in June 2026, according to Flexprice's change log. At 5 credits per second, Gen-4 Turbo keeps drafting costs down before producing final versions with Gen-4.5.

Where it falls short: cost compared with the volume tier, and unused credits on Standard and Pro subscriptions do not carry over.

Client example: No verifiable public client result showing a change in time or cost was found.

Pros:

  • Keyframes, camera controls and an editor built around the model
  • Gen-4 Turbo drafts at 5 credits a second
  • API credits at $0.01 each
  • Partner models such as Veo 3.1 available in the same account

Drawbacks:

  • $1,440 a year for the task, four times the cost of our top choice
  • Subscription credits on Standard and Pro do not carry over

Category scores: Cost 7.2 | Resolution and length 7.6 | Quality 8.8 | Audio and control 7.6 | Access 8.6 | Licensing 7.8

6. Kling 3.0: A leading choice for longer single shots

  • Score: 7.7/10
  • Entry price: Around $0.13 per second ($1.26 per 10-second clip, Atlas Cloud, 2026)
  • Cost for 1,000 seconds per month: Around $1,512 per year
  • Metering: Per second through resellers; subscription credits in Kling's own app
  • Volume threshold: None published
  • Free tier: Limited credits in the consumer app
  • Pricing source: Kling 3.0 via Atlas Cloud, recorded in 2026
Kling 3.0 homepage screenshot

Kling 3.0 gets the job done for around $1,512 a year.

Where it stands out: 10-second single shots. Atlas Cloud lists Kling 3.0 among the few models that generate 10-second clips in a single call.

How it works: Kling sells subscription credits in its own app and offers per-second pricing through resellers. HeyGen's comparison chart lists Kling 3.0 Pro at $0.168 per second at 768p, above the reseller rate.

Where it falls short: cost, with rates varying between Kling's app, its Pro tier and resellers.

Client example: No verifiable published client results showing a change in time or cost were found.

Pros:

  • 10-second clips generated in a single go
  • Photorealistic humans and complex motion in independent coverage
  • Multi-shot storyboard mode
  • Several hosting options

Cons:

  • Around $1,512 a year for the task — 4.2 times the cost of our top choice
  • Pricing varies between the app, the Pro tier and resellers

Category scores: Cost 7.0 | Resolution and length 8.2 | Quality 8.6 | Audio and control 8.2 | Access 7.4 | Licensing 7.0

Tier 3: Open and High-End Models

7. Wan 2.6: The Best Open Model for Draft Videos

  • Score: 7.5/10
  • Entry price: Around $0.07 per second at 720p (Atlas Cloud); $0.10 at 720p and $0.15 at 1080p (ElevenLabs scale-tier reference)
  • Cost for 1,000 seconds per month: Around $840 per year at $0.07
  • Metering: Per second through hosts
  • Volume cliff: 5-second clip cap per Atlas Cloud
  • Free tier: Open weights
  • Pricing source: Wan 2.6 via Atlas Cloud, recorded in 2026
Wan 2.6 homepage screenshot

Wan 2.6 handles the task for around $840 a year when hosted, or just the cost of GPU time if self-hosted.

Its strengths: an open model that is inexpensive for drafting and free to run on your own hardware.

How it works: Wan is Alibaba's family of open models, available through a range of API providers at different rates. With a 5-second limit, a 10-second shot requires two generations.

Where it falls short: clip length and resolution, which is why it scored lowest in that category.

Client example: No verifiable public client result showing a change in time or cost was found.

Pros:

  • Open weights for self-hosting
  • Around $0.07 per second at 720p on Atlas Cloud
  • Many hosting providers
  • Unrestricted prompts on self-hosted runs

Cons:

  • 5-second clips need stitching together for longer shots
  • Hosted rates range from $0.07 to $0.15 per second

Category scores: Cost 8.2 | Resolution and length 6.0 | Quality 7.2 | Audio and control 7.0 | Access 8.0 | Licensing 9.0

8. Veo 3.1: Best for Premium Quality and 4K

  • Score: 7.4/10
  • Entry price: $0.40 per second at 720p to 1080p, $0.60 at 4K (Gemini API, August 2026)
  • Cost for 1,000 seconds a month: $4,800/year
  • Metering: Per second; audio included
  • Volume cliff: None published
  • Free tier: None for video generation
  • Pricing as at: Google Gemini API, captured August 2026
Veo 3.1 Fast homepage screenshot (Google Flow)

Veo 3.1 Standard does the job for $4,800 a year.

Where it stands out: high-end quality, with 4K and audio included in the price.

How it works: the same Gemini billing as Veo 3.1 Fast, at four times the rate. fal offer Veo 3.1 at $0.20 per second without audio and $0.40 with audio, so switching sound off halves the cost on that platform.

Where it falls short: cost, which makes it an expensive choice for high-volume work.

Client example: No verifiable public client result showing a change in time or cost was found.

Pros:

  • Highest documented quality level in the Veo range
  • 4K at $0.60 a second
  • Audio included in Google's rate
  • Independent reviews report close adherence to prompts

Cons:

  • $4,800 a year for the same task, 13.3 times the cost of our top choice
  • 1080p clips are limited to 8 seconds

Category scores: Cost 4.0 | Resolution and length 8.8 | Quality 9.4 | Audio and control 9.0 | Access 8.8 | Licensing 8.0

The Monthly Cost of 1,000 Seconds

The task: 1,000 generated seconds a month at each model's lowest-priced tier offering 720p or above, with sound where included. That is 12,000 seconds a year, or around a hundred 10-second clips a month.

Loading embed content...

Prices from supplier pages and dated third-party snapshots, July to October 2026.

How to check a row: HeyGen Video's standard rate is $0.03 per second at 768p, so 1,000 seconds a month costs $30 and 12,000 seconds a year costs $360. The promotional rate of $0.015, which ended on 31 October 2026, halved those costs.

Resolution affects the ranking. At 1080p, HeyGen Video costs three times its 768p rate, around $0.09 a second, making LTX-2.3 Fast the cheapest model in the table at $0.04 a second.

Where the ranking changes. HeyGen Video is the cheapest option at any volume at 768p. If every clip must be delivered at 1080p, LTX-2.3 is cheapest; if every clip requires Google billing and audio included, Veo 3.1 Lite offers the lowest Veo rate at $0.05 per second at 720p.

Which model costs too much for this task. Veo 3.1 Standard costs $4,800 a year compared with HeyGen Video's $360 — 13.3 times as much — yet scored 7.4 compared with 8.6. The extra $4,440 pays for Veo's highest quality and 4K, which matter more for showcase shots than for producing a hundred clips a month.

AI Video Cost per Second: Specifications Compared

Loading embed content...

Verified against vendor pages and dated captures, July to October 2026.

Loading embed content...

Which Is Right for You?

Based on the work you do

  • If you produce B-roll and product shots in large volumes, use HeyGen Video.
  • If every clip needs to be delivered in 1080p or 4K, use LTX-2.3.
  • If your team works in Google Cloud, use Veo 3.1 Fast.
  • If you edit using keyframes and a timeline, choose Runway Gen-4.5.

By budget

  • Free: LTX-2.3 or Wan 2.6 open weights on your own hardware.
  • Under $50 a month: HeyGen Video at $30 for 1,000 seconds at 768p.
  • Over $300 a month: Veo 3.1, for standout shots where quality matters more than cost.

By how much learning you’re comfortable with

  • I want results today: HeyGen Video through OpenRouter or the HeyGen API with a $5 wallet.
  • I can set aside a weekend: Runway Gen-4.5, whose editor rewards practice.
  • I have a team to train: LTX-2.3, where self-hosting becomes cost-effective at scale.

Too New to Assess

LTX-2.5 launched on 11 August 2026, with Fast priced at $0.09 per second at 720p and $0.13 at 1080p via the official API. It will enter the ranking once independent reviews establish how its quality compares with LTX-2.3. I will review it again in January 2027.

Veo 3.1 Lite starts at $0.05 a second at 720p on the Gemini API. It would enter Tier 1 once its quality compared with Veo 3.1 Fast has been independently assessed. I will review it again in January 2027.

How to Choose: A Model Priced per Second or a Platform Priced per Minute

A per-second model generates scenes: B-roll, product shots, and stylised sequences of 5 to 15 seconds. A per-minute avatar platform provides a presenter who reads your script. Teams that need both usually pay for both, so cost each job in its own unit: seconds for scenes, minutes for presenters.

For scenes, start with the per-second table above. For presenter videos, compare per-minute platforms such as HeyGen's AI video generator, where avatar minutes are billed separately from generated scenes.

Frequently asked questions

How much does AI video cost per second in 2026?

From about $0.03 to $0.40. As of October 2026, listed prices are $0.03 at 768p for HeyGen Video, $0.04 at 1080p for LTX-2.3 Fast, $0.10 at 720p for Veo 3.1 Fast and $0.40 for Veo 3.1 Standard.

Which AI video model has the lowest cost per second?

HeyGen Video at 768p. It costs $0.03 a second at the standard rate; a $0.015 launch promotion ended on 31 October 2026. At 1080p, LTX-2.3 Fast at $0.04 is cheaper.

How much does Veo 3.1 cost per second?

$0.40 at 720p to 1080p and $0.60 at 4K on Google's Gemini API, with audio included. Veo 3.1 Fast costs $0.10 at 720p and Lite $0.05.

Which costs less per second: Kling or Runway?

Runway, by a small margin. Runway Gen-4.5 costs $0.12 per second through the API, compared with around $0.13 for Kling 3.0 through Atlas Cloud. Kling is the better choice if you need 10-second single shots.

Can I still use Sora through an API?

No. OpenAI closed the Sora app on 26 April 2026 and discontinued the Sora API on 24 September 2026.

What mistake do people most often make when comparing AI video pricing?

Comparing headline prices for different resolutions. 480p and 1080p are different products, so compare every model at the resolution you deliver.

Does audio cost extra?

It depends on the host. Google include audio in Veo 3.1 rates, while fal charge $0.20 per second for Veo 3.1 with audio off and $0.40 with audio on.

The verdict

HeyGen Video has the lowest cost per second in this comparison: $0.03 at 768p with native sound, bringing the cost of 1,000 seconds a month to $360 a year. LTX-2.3 is the best choice for 1080p and above. Veo 3.1 Fast suits teams using Google Cloud. Veo 3.1 Standard is suited to showcase shots. Start with $5 in your HeyGen API wallet and compare the cost of your own clips with the table.

About

Greetings! My name is Ayesha Shaheryar. My words have helped millions over the past two years. As a HeyGen expert and a writer, I am here to introduce tips and tricks to edit your next video in no time.

Continue Reading

Latest blog posts related to AI Video Costs per Second in 2026: Comparing 8 Models on a Real Project.

Browse All

Start creating videos with AI

See how businesses like yours scale content creation and drive growth with the most innovative AI video.