Field dossier - Moving pictures

Top AI video generators.

Matt Farmer's current ranked shortlist of the top AI video generators: practical output quality, motion, realism, multimedia reference control, audio, and creator workflow value without turning the page into a spec dump.

Updated June 28, 2026Ranked by practical outputSeedance 2.5 on watchlist

Top five comparison

Fast read before the deep dive.

Specs move quickly, so this grid favors decision signals over exhaustive API detail. Verify current pricing and endpoint availability before production use.

Model

Max duration

Resolution

Native audio

Multimedia refs

Best use

Seedance 2.0

Max duration

Up to 15 sec in common integrations

Resolution

480p, 720p, 1080p, and native 4K

Native audio

Generated audio support

Multimedia refs

Multimedia refs: text, image, audio, video

Best use

High-end cinematic AI video, product shots, and reference-heavy creative work.

Veo 3.1

Max duration

Commonly 4, 6, or 8 sec

Resolution

720p, 1080p, and 4K/upscale paths

Native audio

Native audio and dialogue

Multimedia refs

Ingredients, reference images, first/last frame

Best use

Realistic short scenes, physics-heavy clips, and native-audio experiments.

Kling 3.0

Max duration

Up to 15 sec in current high-end workflows

Resolution

720p, 1080p, and 4K paths

Native audio

Native audio in supported modes

Multimedia refs

Omni refs: images, video, elements, start/end frames

Best use

Directed scenes, storyboard-style control, character consistency, and reference-heavy workflows.

Grok Imagine 1.5

Max duration

1 to 15 sec in current docs

Resolution

480p, 720p, and 1080p tiers

Native audio

Synchronized audio experience

Multimedia refs

Image-to-video first-frame workflow

Best use

Fast image-to-video experiments, surreal social clips, and attention-grabbing creative tests.

Vidu Q3

Max duration

Up to 16 sec for Q3 Pro/Turbo

Resolution

540p, 720p, and 1080p

Native audio

Synchronized audio-video output

Multimedia refs

Text, image, start/end frame, subject refs

Best use

Longer clips, audio-video output, reference workflows, and budget-conscious production tests.

Ranked field report

Dreamina Seedance 2.0

High-end cinematic AI video

The strongest current pick when the brief needs beauty, motion, physics, and multimodal references in the same lane.

Strengths

Cinematic motion and visual polish
Text, image, audio, and video reference workflows
Strong fit for hero shots and social creative

Tradeoffs

Access and pricing can vary by platform
Seedance 2.5 claims still need public production verification

Max duration

Up to 15 sec in common integrations

Resolution

480p, 720p, 1080p, and native 4K

Native audio

Generated audio support

Multimedia refs

Multimedia refs: text, image, audio, video

Dreamina by CapCut Watch launch trailer

Realism

Motion

Control

Audio

Value

Best realism

Google Veo 3.1

Realistic short scenes

The cleanest realism pick: strong physics, natural camera movement, good prompt following, and official Google docs.

Strengths

Excellent realism and lighting
Native audio, dialogue, and sound effects
Clear Gemini API documentation and pricing paths

Tradeoffs

Shorter clip lengths than several competitors
4K availability depends on product and endpoint context

Max duration

Commonly 4, 6, or 8 sec

Resolution

720p, 1080p, and 4K/upscale paths

Native audio

Native audio and dialogue

Multimedia refs

Ingredients, reference images, first/last frame

Gemini Watch DeepMind video DeepMind Veo 3.1 Google Flow

Realism

Motion

Control

Audio

Value

Best control

Kling 3.0

Directed scenes

Kling has moved into top-tier territory because it is becoming more director-like, especially around references and story control.

Strengths

Strong reference and storyboard workflows
Character consistency and multi-shot potential
4K support in current high-end paths
Native audio and dialogue-style capabilities

Tradeoffs

Credit costs, 4K pricing, and mode availability can be harder to compare
Dialogue still needs careful prompting or post-production

Max duration

Up to 15 sec in current high-end workflows

Resolution

720p, 1080p, and 4K paths

Native audio

Native audio in supported modes

Multimedia refs

Omni refs: images, video, elements, start/end frames

Kling AI Watch Kling video

Realism

Motion

Control

Audio

Value

Best social spark

Grok Imagine 1.5

Fast image-to-video experiments

Not the most polished frontier model yet, but fast, fun, social-native, and improving quickly.

Strengths

Fast image-to-video iteration
Strong weird, fun, and surreal creative range
Straightforward API model docs

Tradeoffs

Grok Imagine Video 1.5 is image-to-video only
Less consistently polished than Seedance, Veo, or Kling

Max duration

1 to 15 sec in current docs

Resolution

480p, 720p, and 1080p tiers

Native audio

Synchronized audio experience

Multimedia refs

Image-to-video first-frame workflow

Grok Imagine Watch Grok video xAI 1.5 notes xAI model docs

Realism

Motion

Control

Audio

Value

Best practical value

Vidu

Longer clips

The underrated practical pick: Q3 Pro and Q3 Turbo make Vidu useful when you need more than a short 8-second test.

Strengths

Up to 16-second Q3 clips
Audio-video generation and subject references
Good balance of quality, duration, and price

Tradeoffs

Less awareness than Google, Runway, Kling, or xAI
May not beat Seedance or Veo on pure beauty or realism

Max duration

Up to 16 sec for Q3 Pro/Turbo

Resolution

540p, 720p, and 1080p

Native audio

Synchronized audio-video output

Multimedia refs

Text, image, start/end frame, subject refs

Vidu Official Q3 deep dive

Upcoming model

Watchlist, not a rank.

Announced models can reshape a list overnight. Seedance 2.5 gets a clear watchlist slot here, but the ranking stays focused on models with enough public production signal.

Announced watchlist

Seedance 2.5

Seedance 2.5 is the model to watch next. It belongs on the page, but not in the ranked list until access, pricing, docs, and real-world output are production-verified.

Reported native 30-second clipsReported native 4K generationReported support for many multimodal references

Watch official updates

Wider landscape

Not ranked, still worth knowing.

These tools and models are not numbered because the page should stay decisive. They still matter for specific workflows like lip sync, open pipelines, mature creator tools, or cheap iteration.

Audio-driven workflows

Wan 2.7

Strong for custom audio, first-and-last-frame control, video continuation, and music-video-style work.

AudioControlNarrative

Cinematic creative tests

HappyHorse 1.1

A rising Alibaba model worth testing for visual quality, dynamic motion, and discounted creator experiments.

Cinematic15 secAlibaba

Camera motion and value

MiniMax Hailuo 2.3

Useful for image-to-video, physics improvements, expressive animation, and camera command workflows.

I2VPhysicsValue

Polished creator workflows

Runway Gen-4.5

Still important because Runway has a mature interface, strong creative tooling, and a serious API ecosystem.

PlatformAPIWorkflow

Builder and pipeline work

LTX-2.3

A practical option for open/local experimentation, synchronized audio-video, retake, extend, and 4K workflows.

Open4KPipeline

Pricing field guide

What each AI video model actually costs.

Pricing is normalized around cost per second whenever the vendor publishes enough detail. Credit-only rows stay credit-only unless the platform gives a clean USD conversion, and creator platforms without clean API-style pricing are left out.

1080p quick comparison

Main and cheapest cost per generation.

Start with each provider's main model tier, then switch to the cheapest versions tab for Mini, Lite, Turbo, Fast, or no-audio lanes. The table uses 10 seconds at 1080p unless a platform only publishes a shorter 1080p maximum.

Main model order

Costs use 10 seconds at 1080p unless a platform only publishes a shorter 1080p max.

Model	Platform	Example length	1080p cost/sec	Example generation	Source
Seedance 2.0	Segmind	10 sec	$0.34/sec	$3.40 Segmind text/image-to-video 1080p rate; 10 seconds is derived from the per-second price.	Segmind Seedance pricing
Google Veo 3.1	Gemini / Vertex AI	8 sec	$0.40/sec	$3.20 Standard 1080p with audio. Video-only is lower.	Google pricing
Kling VIDEO 3.0	Kling AI	10 sec	~$0.16/sec	~$1.60 Based on the lowest paid package at $1.33 per 100 credits and Kling's 12 credits/sec 1080p native-audio rate. No-audio is about $0.11/sec.	Kling guide
Grok Imagine Video 1.5	xAI API	10 sec	$0.25/sec	$2.50 Image-to-video route; check the source page for current input-media billing details.	xAI pricing
Vidu Q3 Pro	Vidu API	10 sec	$0.12/sec	$1.20 Main Pro tier; Turbo is cheaper for drafts.	Vidu pricing
Wan 2.7	Alibaba Cloud	10 sec	$0.15/sec	$1.50 International Model Studio pricing.	Alibaba pricing
HappyHorse 1.1	Alibaba Cloud	10 sec	$0.18/sec	$1.80 International Model Studio pricing.	Alibaba pricing
MiniMax Hailuo 2.3	MiniMax API	6 sec	~$0.082/sec	$0.49 MiniMax lists Hailuo 2.3 at $0.49 per 1080p, 6-second video; the per-second number is $0.49 / 6.	MiniMax pricing
Runway Gen-4.5	Runway API	10 sec	$0.12/sec	$1.20 Runway's own model pricing, not a third-party wrapper row.	Runway pricing
LTX 2.3 Pro	LTX API	10 sec	$0.08/sec	$0.80 Pro 1080p tier; Fast is cheaper.	LTX pricing

Kling estimate uses the current lowest paid package baseline: 660 credits/month advertised as low as $1.33 per 100 credits. The main tab uses the 1080p native-audio rate of 12 credits per second; the cheapest tab uses the 1080p no-audio rate of 8 credits per second.

Cheapest clean API lanes

Vidu Q3 Turbo off-peak, Veo 3.1 Lite video-only, Hailuo 2.3 Fast, Runway Gen-4 Turbo, and LTX Fast are the low-cost iteration paths.

Most expensive final tier

Seedance 2.0 native 4K through Segmind is materially higher than its 1080p row, so compare resolution before production spend.

Cleanest pricing

Segmind, Google, xAI, Vidu, Alibaba, MiniMax, LTX, and Runway publish enough detail to normalize the math.

Hardest to compare

Kling is credit-based, and Seedance video-to-video can vary because reference media changes the token load.

ByteDance

Seedance 2.0 via Segmind

Dreamina is the preferred consumer surface, while Segmind is the cleaner current API pricing path for Seedance 2.0.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
Seedance 2.0	Segmind	Text / image-to-video	480p	$0.0703/sec	Token-based, shown as USD per generated second	Native audio included	Lowest-resolution draft lane for text prompts or reference images.	Segmind pricing
Seedance 2.0 Mini	Segmind	Mini text / image-to-video	720p	~$0.0756/sec	Derived from Segmind's Mini half-cost rule	Native audio included	Mini is the cheaper draft lane, but Segmind's Mini pricing page frames it around 480p/720p rather than the 1080p comparison target.	Segmind pricing
Seedance 2.0	Segmind	Text / image-to-video	720p	$0.1512/sec	Token-based, shown as USD per generated second	Native audio included	Strong balance for social drafts and practical review work.	Segmind pricing
Seedance 2.0	Segmind	Text / image-to-video	1080p	$0.34/sec	Token-based, shown as USD per generated second	Native audio included	This is the main 1080p Seedance comparison row for API pricing.	Segmind pricing
Seedance 2.0	Segmind	Text / image-to-video	4K	$1.3721/sec	Token-based, shown as USD per generated second	Native audio included	Use when native 4K is worth the higher render cost.	Segmind pricing
Seedance 2.0	Segmind	Video-to-video typical	1080p	~$0.41/sec	Typical generated-output cost; reference video changes token count	Native audio included	Segmind lists 1080p video-to-video as a typical range around $0.35-$0.50/sec.	Segmind pricing

Google

Veo 3.1 via Gemini / Vertex AI

Google has the cleanest pricing ladder because Veo 3.1 is split by model speed, audio, and resolution.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
Veo 3.1	Gemini / Vertex AI	Standard + audio	720p / 1080p; 4K	$0.40/sec; $0.60/sec	USD per generated second	Native speech and sound effects	Best realism path when you want the full audio-video model.	Google pricing
Veo 3.1	Gemini / Vertex AI	Standard, video only	720p / 1080p; 4K	$0.20/sec; $0.40/sec	USD per generated second	No native audio	A better comparison when audio will be added in post.	Google pricing
Veo 3.1 Fast	Gemini / Vertex AI	Fast + audio	720p / 1080p / 4K	$0.10 / $0.12 / $0.30/sec	USD per generated second	Native speech and sound effects	Strong draft or social-testing lane if quality is good enough.	Google pricing
Veo 3.1 Fast	Gemini / Vertex AI	Fast, video only	720p / 1080p / 4K	$0.08 / $0.10 / $0.25/sec	USD per generated second	No native audio	Clean fast route when you care more about picture than native sound.	Google pricing
Veo 3.1 Lite	Gemini / Vertex AI	Lite + audio	720p / 1080p	$0.05 / $0.08/sec	USD per generated second	Native speech and sound effects	Cheapest direct Veo route with audio, but no 4K tier listed.	Google pricing
Veo 3.1 Lite	Gemini / Vertex AI	Lite, video only	720p / 1080p	$0.03 / $0.05/sec	USD per generated second	No native audio	The cheapest published direct Veo option for rough iteration.	Google pricing

Kuaishou

Kling 3.0 / Kling 3.0 Omni

Kling publishes credit-per-second math, but the USD equivalent depends on the exact credit package, region, product surface, and 4K availability path.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
Kling VIDEO 3.0	Kling AI	No native audio	720p / 1080p	Credit-based	6 / 8 credits/sec	No native audio	Base generation price before audio or voice-control add-ons.	Kling guide
Kling VIDEO 3.0	Kling AI	Native audio	720p / 1080p	Credit-based	9 / 12 credits/sec	Native audio on	Best apples-to-apples row against Veo or Vidu native-audio outputs.	Kling guide
Kling VIDEO 3.0	Kling AI	Native 4K	4K	Credit-based	Listed on Kling dev pricing; verify account rate	Mode dependent	Kling now supports 4K, but the public guide exposes cleaner 720p/1080p credit examples than 4K math.	Kling dev pricing
Kling VIDEO 3.0	Kling AI	Voice control add-on	720p / 1080p / 4K paths	Credit-based	+2 credits/sec	Voice tone control	Add this on top of native-audio generation when controlling voice tone.	Kling guide
Kling VIDEO 3.0 Omni	Kling AI	No video input	720p / 1080p	Credit-based	6 / 8 credits/sec no audio; 9 / 12 credits/sec audio	Optional native audio	Omni's all-in-one reference workflow covers text, image, video, audio, and element context.	Kling Omni guide
Kling VIDEO 3.0 Omni	Kling AI	With video input	720p / 1080p	Credit-based	12 / 16 credits/sec	Native audio not supported yet with video input	This is where Omni reference/video-input workflows get more expensive.	Kling Omni guide

xAI

Grok Imagine

xAI is straightforward: Grok Imagine Video 1.5 is the pricier image-to-video model, while the broader Grok Imagine Video route is cheaper but lower-resolution.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
Grok Imagine Video 1.5	xAI API	Image-to-video	480p / 720p / 1080p	$0.08 / $0.14 / $0.25/sec	USD per output second; input media billed separately	Synchronized audio experience	Use this row for the current higher-end Grok Imagine 1.5 comparison.	xAI pricing
Grok Imagine Video	xAI API	Text, image, or video to video	480p / 720p	$0.05 / $0.07/sec	USD per output second; input media billed separately	Synchronized audio experience	Cheaper and more flexible, but not the same as the 1.5 image-to-video route.	xAI pricing

ShengShu

Vidu Q3

Vidu is one of the clearest value stories: Q3 Turbo is cheap, Q3 Pro is still reasonable, and Q3 Mix costs more for reference-to-video.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
Vidu Q3 Pro	Vidu API	T2V / I2V / start-end	540p / 720p / 1080p	$0.045 / $0.10 / $0.12/sec	9 / 20 / 24 credits/sec	Audio-video generation supported	Off-peak drops to $0.025 / $0.05 / $0.06 per second.	Vidu pricing
Vidu Q3 Turbo	Vidu API	T2V / I2V / start-end	540p / 720p / 1080p	$0.035 / $0.055 / $0.065/sec	7 / 11 / 13 credits/sec	Audio-video generation supported	Off-peak drops to $0.02 / $0.03 / $0.035 per second.	Vidu pricing
Vidu Q3 Mix	Vidu API	Reference-to-video	720p / 1080p	$0.12 / $0.145/sec	24 / 29 credits/sec	Reference workflow	No off-peak discount is listed for this Mix row.	Vidu pricing

Alibaba

Wan 2.7 and HappyHorse 1.1

Alibaba's international pricing is cleanly published in USD per output second for both Wan and HappyHorse.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
Wan 2.7	Alibaba Cloud Model Studio	T2V / I2V / R2V	720p / 1080p	$0.10 / $0.15/sec	USD per output second	Audio video	International Singapore pricing, with 50 seconds of free quota listed.	Alibaba pricing
HappyHorse 1.1	Alibaba Cloud Model Studio	T2V / I2V / R2V	720p / 1080p	$0.14 / $0.18/sec	USD per output second	Audio video	International Singapore pricing, with 10 seconds of free quota listed.	Alibaba pricing

MiniMax

Hailuo 2.3

MiniMax publishes per-generation pricing, so the useful normalized number is approximate cost per second by duration and resolution.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
MiniMax Hailuo 2.3 Fast	MiniMax API	Fast	768p, 6s / 10s	$0.19 / $0.32 per clip	Approx. $0.032/sec	Not primary strength	One of the cheapest clean API iteration paths in the table.	MiniMax pricing
MiniMax Hailuo 2.3 Fast	MiniMax API	Fast	1080p, 6s	$0.33 per clip	Approx. $0.055/sec	Not primary strength	Fast 1080p is still cheaper than many premium 1080p routes.	MiniMax pricing
MiniMax Hailuo 2.3	MiniMax API	Standard	768p, 6s / 10s	$0.28 / $0.56 per clip	Approx. $0.047-$0.056/sec	Not primary strength	More expensive than Fast, but still practical for testing.	MiniMax pricing
MiniMax Hailuo 2.3	MiniMax API	Standard	1080p, 6s	$0.49 per clip	$0.49 per 1080p, 6s video	Not primary strength	Normalized to about $0.082/sec by dividing the 6-second clip price.	MiniMax pricing

Runway

Gen-4.5 and Gen-4 Turbo

Runway remains a polished creator/API platform, and its own model pricing is simple in credits per generated second.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
Runway Gen-4.5	Runway API	Premium Runway model	Model-defined output	$0.12/sec	12 credits/sec	Not primary strength	Useful if you want Runway's platform and API ecosystem, not the cheapest base model.	Runway pricing
Runway Gen-4 Turbo	Runway API	Fast Runway model	Model-defined output	$0.05/sec	5 credits/sec	Not primary strength	Cheaper Runway-native iteration route.	Runway pricing

Lightricks

LTX 2.3

LTX pricing is clean and useful for builders because it separates Fast, Pro, 1080p, 1440p, and 4K.

Model	Platform	Mode	Resolution	Cost	Billing	Audio	Read this as	Source
LTX 2.3 Fast	LTX API	T2V / I2V	1080p / 1440p / 4K	$0.06 / $0.12 / $0.24/sec	USD per generated second	Synchronized audio workflows	Good builder route when 4K or pipeline control matters.	LTX pricing
LTX 2.3 Pro	LTX API	T2V / I2V	1080p / 1440p / 4K	$0.08 / $0.16 / $0.32/sec	USD per generated second	Synchronized audio workflows	Higher fidelity route while staying cheaper than many premium 4K options.	LTX pricing
LTX 2.3 Pro	LTX API	Audio-to-video / retake / extend	1080p	$0.10/sec	USD per input or generated second by route	Audio-conditioned workflows	Useful for pipeline edits and audio-driven work rather than pure prompt tests.	LTX pricing

Rates can change without warning. Platform wrapper pricing is not always the same as first-party API pricing, and native 4K is not the same thing as an upscale. Use this table as a practical starting point, then verify the linked pricing page before production spend.

Methodology

Verified specs plus Matt's working opinion.

Verified facts

Model pages, API docs, pricing pages, and official release notes are used for names, modes, durations, and availability signals.

Practical ranking

The order is based on creator usefulness: output quality, motion, reference handling, audio, workflow fit, and value.

Fast-moving market

AI video specs change quickly. Treat this as a dated field report and verify exact pricing or API behavior before production.

Work with Matt

Need help choosing the right AI tools?

Book a strategy hour with Matt for practical help picking AI video tools, building a workflow, comparing paid options, or figuring out how AI fits into your business.

Work with Matt

Top AI video generators.

Fast read before the deep dive.

Seedance 2.0

Veo 3.1

Kling 3.0

Grok Imagine 1.5

Vidu Q3

Top 5 AI video generators.

Dreamina Seedance 2.0

Strengths

Tradeoffs

Google Veo 3.1

Strengths

Tradeoffs

Kling 3.0

Strengths

Tradeoffs

Grok Imagine 1.5

Strengths

Tradeoffs

Vidu

Strengths

Tradeoffs

Watchlist, not a rank.

Seedance 2.5

Not ranked, still worth knowing.

Wan 2.7

HappyHorse 1.1

MiniMax Hailuo 2.3

Runway Gen-4.5

LTX-2.3

What each AI video model actually costs.

Main and cheapest cost per generation.

Cheapest clean API lanes

Most expensive final tier

Cleanest pricing

Hardest to compare

Seedance 2.0 via Segmind

Veo 3.1 via Gemini / Vertex AI

Kling 3.0 / Kling 3.0 Omni

Grok Imagine

Vidu Q3

Wan 2.7 and HappyHorse 1.1

Hailuo 2.3

Gen-4.5 and Gen-4 Turbo

LTX 2.3

Verified specs plus Matt's working opinion.

Verified facts

Practical ranking

Fast-moving market

Need help choosing the right AI tools?