Cycling through all six. Tap any point to stop.

Measured on six things

AI that makes 1080p multi-shot videos with synced sound.

Speed

Desktop Lighthouse 94/100 but mobile collapses to 51/100 with LCP 5.9s, CLS 0.249 and TBT 501ms — a heavy mobile experience for a tool people will open on phones to make 9:16 clips.

Screenshots

1 / 1

Wan 2.6 Video Generator

About this app

Wan 2.6 Video Generator is a browser-based AI video creation tool that turns text prompts, reference images, or short ref clips into 1080p, 15-second multi-shot videos with natively synced audio and lip-sync. I spent a few afternoons poking at it, and what genuinely stands out is the multi-shot storytelling—you can plan a sequence with smooth camera pans and close-ups, and the model keeps faces, lighting, and voices consistent across shots. The reference-to-video mode is the sleeper feature: feed it a 5-second clip and it replicates a character's face and voice for one or two people, which is huge for recurring social content. Output is 24fps at 1080p across 1:1, 16:9, and 9:16, renders average under 30 seconds, and you get full commercial rights. Pricing is one-time credit packs (Starter ~$99.90/yr for ~120 videos up to Premium ~$399.90/yr with API access) and credits never expire, which I appreciate versus yet another subscription. There's a free tier with starter credits and no c...

Key features

1Multi-shot 1080p video generation with consistent faces, lighting and voices across shots
2Reference-to-video mode that replicates a character's face and voice from a ~5-second clip
3Natively synced audio and lip-sync at 24fps in 1:1, 16:9 and 9:16 aspect ratios
4One-time credit packs (~$99.90–$399.90/yr) with credits that never expire, plus a free tier
5Full commercial rights on generated output; API access on the Premium tier

Target customer profile

Drawn from the product itself, not from a survey.

Demographic

Content creators & social media managers (TikTok, Reels, Shorts)

Pain points

Editing skills lacking; slow manual production; need consistent faces across clips

Primary needs

Fast 9:16 short clips with sync audio and steady characters

Demographic

Indie filmmakers, motion designers & video producers

Pain points

Expensive software; slow concept visualization; character consistency across shots

Primary needs

Multi-shot cinematic sequences with reference-to-video replication

Demographic

Marketers, educators & small business teams

Pain points

Tight budgets; no video team; confusing or pricey tools; licensing worries

Primary needs

Commercial-use 1080p videos from simple prompts and images

Scout Review

Written by AI from measured evidence, scored out of 100.

Wan 2.6 Video Generator has a genuinely interesting pitch — 1080p, 15-second multi-shot video with synced audio, lip-sync and character consistency from a short reference clip — sold through non-expiring credit packs instead of yet another subscription. The generator widget is well organized and pricing is transparent at the point of action. But the execution undercuts the promise: the landing page shows zero actual video output, the stats strip is missing its numbers, mobile performance is poor (Lighthouse 51, LCP 5.9s), security headers are largely absent, and accessibility fails on contrast, button names and touch targets. It's a promising engine wrapped in a marketing page that forgot to show the engine running.

Measure by measure

Usability

Generator groups prompt, reference, duration, ratio and cost sensibly and shows credit cost at the button, but the reference image is '(Required)' in the tab and '(optional)' in the dropzone, and example prompts are tiny grey text.

Design

Coherent dark palette with one purple accent and a clean generator widget, but the page never shows a single video — every demo tile is an empty black box and the stats strip has labels with no numbers.

Speed

Desktop Lighthouse 94/100 but mobile collapses to 51/100 with LCP 5.9s, CLS 0.249 and TBT 501ms — a heavy mobile experience for a tool people will open on phones to make 9:16 clips.

Security

HTTPS enforced with SPF and DMARC, but no HSTS, CSP, frame protection, Referrer-Policy or secure cookies (4/13 checklist points). No SOC 2, ISO, pen-test or bug bounty surfaced in collected evidence.

Accessibility

Lighthouse a11y 87/100 but fails on unnamed buttons, insufficient contrast and undersized touch targets; English-only with no WCAG statement. The low-contrast grey body text confirms the contrast failure.

Growth

Real, specific wedge: multi-shot consistency plus reference-to-video character replication, sold as non-expiring credit packs against subscription rivals. Early (domain ~0.9yr) but the positioning is concrete.

Improvement areas

Landing page shows no actual video output — every demo tile is an empty placeholder
Mobile performance is weak (Lighthouse 51, LCP 5.9s, CLS 0.249) for a phone-first audience
Security headers largely missing (4/13) with no published data-handling or compliance page
Accessibility fails on button names, color contrast and touch target size; English-only

Design & Usability

The design system is genuinely coherent — dark canvas, one purple accent, a single white CTA, and a generator widget that groups mode tabs, prompt, reference upload, duration, aspect ratio and credit cost in a logical order. The problem is the product never shows itself: every 'Create Like This' tile and feature card is an empty black placeholder, the hero's right half is dead space, and the stats strip prints labels like 'Platform uptime' with no figures. For a tool whose entire value is motion, that's a self-inflicted wound. Usability is competent where it counts — cost is visible right above the generate button and there's a no-credit-card free tier — but the reference image is '(Required)' in the tab and '(optional)' in the dropzone, and the example prompts that should onboard newcomers are tiny grey text in a scroll box.

Speed & Security

Speed is a tale of two devices: desktop Lighthouse lands at 94/100, but mobile drops to 51/100 with LCP 5.9s, CLS 0.249 and TBT 501ms. Since the stated audience is TikTok/Reels/Shorts creators working in 9:16, the mobile number is the one that matters, and it's weak. Security is thin: HTTPS is enforced and SPF/DMARC are set, but there's no HSTS, CSP, frame protection, Referrer-Policy or secure cookies — 4 of 13 checklist points. No SOC 2, ISO, pen-test or bug bounty page surfaced. The site promises 'projects stay private' and commercial rights, but publishes no data-handling documentation to back the privacy claim, which matters when users upload reference clips of real faces.

Accessibility & Growth

Accessibility scores 87/100 on Lighthouse but fails on three concrete audits: buttons without accessible names, insufficient color contrast, and touch targets too small. The visual audit independently confirms the contrast problem — body copy is faint grey on near-black. The site is English-only and publishes no WCAG statement. Growth is the strongest story here: the wedge is real and specific. Multi-shot storytelling with consistent faces, lighting and voices, plus reference-to-video replication of a character from a 5-second clip, is exactly what serial social content needs, and one-time credit packs that never expire are a genuine differentiator against Runway, Pika and Kling's subscriptions. The category has headroom, but the site's own numbers (75K+ creators, 3,066 videos/24h) are unverified and the domain is under a year old.

Conclusion

If you're a creator who needs consistent characters across a series of 9:16 clips and you're tired of subscription lock-in, the credit-pack model and reference-to-video mode are worth a free-tier test. Go in expecting to judge the output yourself, because the site won't show it to you. Before uploading reference clips of real people, ask the team directly how that footage is stored and deleted — the privacy claim is currently just a sentence. The bones are here; the presentation and the hardening are not yet.

Suggestions for improvement

Put real generated videos in every demo tile and the hero — an empty black placeholder on a video product is the single biggest conversion killer.
Fix the mobile Lighthouse score: target LCP under 2.5s and CLS under 0.1, since your audience works on phones.
Resolve the '(Required)' vs '(optional)' reference-image contradiction and enlarge the example prompts into scannable cards.
Add HSTS, CSP, frame protection and a short security/data-handling page covering how uploaded reference clips are stored.

How we score

Alternatives comparison

Named competitors, point by point. Nobody paid to appear here or to be left out.

	Wan 2.6 Video Generator	Runway	Pika	Kling AI
Pricing model	One-time credit packs (~$99.90–$399.90/yr), credits never expire, free tier	Subscription-based tiers with monthly credit allowances	Subscription with its own credit system	Credit-based plans via web app, subscription tiers
Max clip length / resolution	15s, 1080p, 24fps, multi-shot	Short clips extendable via editing tools; up to 4K on higher tiers	Short social-length clips, typically a few seconds per generation	Known for longer clips and strong motion quality
Native audio / lip-sync	Natively synced audio and lip-sync claimed	Audio tools available; lip-sync via separate features	Limited native audio; effects-focused	Audio/lip-sync support varies by model version
Character consistency across shots	Reference-to-video replicates face and voice from a 5s clip	Character reference and consistency tools available	Limited character consistency features	Reference-based consistency on newer models
Commercial usage rights	Full commercial rights claimed on all output	Commercial use on paid plans	Commercial use on paid tiers	Commercial use per plan terms