Kling AI
AI video generator that creates 4K clips up to 15 seconds from text or image prompts
Alibaba's open-source AI video model family (Wan 2.1–2.7) with a hosted platform at wan.video for text-to-video and image-to-video generation
Wan is Alibaba Tongyi Lab's open-source video generation model family, hosted for browser-based use at wan.video. The hosted platform runs on credits: a limited free tier covers testing, Pro costs $5/month (billed yearly) for 300 credits, and Premium is $20/month (billed yearly) for 1,200 credits — roughly 60 to 240 standard videos monthly. The underlying model weights are Apache 2.0 licensed and free to self-host. Best for creators who want a cheap hosted option and developers who want an open model.
Wan is a family of open-source text-to-video and image-to-video generation models built by Alibaba’s Tongyi Lab, released under the Apache 2.0 license with model weights, training code, and inference scripts publicly available. The model line has moved fast: Wan 2.1 established the base text/image-to-video capability, Wan 2.2 added a 27-billion-parameter mixture-of-experts architecture and character animation, Wan 2.5 (September 2025) introduced native audio — generating synchronized dialogue, sound effects, and lip movement in the same pass as the video — and Wan 2.6 (December 2025) added multi-shot storytelling and consistent character identity across scenes using up to 150 reference frames. By July 2026, Wan 2.7 is the latest release in the family.
Alongside the open-source weights, Alibaba runs a hosted consumer platform at wan.video (the generation interface lives at create.wan.video) where anyone can generate text-to-video and image-to-video clips up to 1080p at 24fps in a browser, without setting up GPU infrastructure. The hosted platform runs on a credit system: a limited free tier lets you test generation, and paid Pro and Premium memberships unlock higher monthly credit allotments and commercial usage rights. Developers who want programmatic access instead of the consumer interface can call Wan models through Alibaba Cloud’s Model Studio (DashScope) API, paying per clip rather than a monthly credit allowance.
Wan suits two fairly different audiences: people who just want to generate a video from a prompt in a browser without deep technical setup, and developers or studios who want an open-weight model they can self-host, fine-tune, or embed in their own pipeline without per-video licensing fees.
Starting price: $5/mo · Free tier: yes · Model: freemium
Price history tracked from June 2026
| Plan | Price | Includes |
|---|---|---|
| Free | Free | 1 concurrent video and 1 concurrent image generation job · 6 image styles · Testing use only, no commercial license |
| Pro | $5/mo | 300 credits/month (roughly 60 standard videos) · Priority generation queue · Commercial use rights |
| Premium | $20/mo | 1,200 credits/month (roughly 240 standard videos) · Priority generation queue · Commercial use rights |
| Credit packs | $1.50–$100 | 30 credits for $1.50 up to 3,900 credits for $100 · No subscription required · 1-year redemption validity |
| API (Alibaba Cloud Model Studio / DashScope) | $0.10–$0.15 | Official Alibaba Cloud API access to Wan models · Pay-per-generation, no subscription · Separate product from the wan.video consumer platform |
| Pros | Cons |
|---|---|
| Model weights are Apache 2.0 licensed — free commercial use and no per-video royalties for teams willing to self-host | The official wan.video pricing page renders via JavaScript, so plan details aren't visible in a static page fetch — pricing here is corroborated from third-party guides rather than a direct read of the live page |
| Rapid version cadence (2.1 through 2.7 released within about two years) has kept quality and features competitive with closed rivals | Advertised $5 and $20 monthly rates on Pro and Premium require yearly billing; standard month-to-month pricing isn't clearly published |
| Cheapest paid hosted membership starts at $5/month (billed yearly) for 300 credits, undercutting most closed-source video generators | Free tier is capped at 1 concurrent video/image job and is limited to testing, not commercial output |
| Native audio-video sync (dialogue, sound effects, lip movement) since Wan 2.5 removes the need for a separate audio pipeline | The Wan name is heavily mirrored by unofficial third-party sites (e.g. wan2-1.com, wan27.org) running their own pricing and free-credit offers, which makes it easy to land on the wrong site |
AI video generator that creates 4K clips up to 15 seconds from text or image prompts
AI video generator that creates cinematic clips from text or images using the Ray 2 model
AI video generation platform for cinematic clips, world simulation, and conversational video agents
AI video generator that turns text prompts and photos into short-form videos with scene editing tools
AI video generator by MiniMax that turns text and images into short cinematic clips at up to 1080p
It depends on how you use it. The underlying model weights are Apache 2.0 licensed and free to self-host with no royalties. On the hosted wan.video platform, there's a limited free tier (1 concurrent video/image job, testing only); real usage needs a paid Pro ($5/month, billed yearly) or Premium ($20/month, billed yearly) membership, or a one-time credit pack.
Wan is developed by Alibaba's Tongyi Lab. It's released as an open-source model family alongside a consumer-facing hosted platform at wan.video.
Wan is the open-source model family (weights, training code, and inference scripts under Apache 2.0) that anyone can self-host. wan.video is Alibaba's own hosted platform that runs those models behind a browser UI and a credit-based subscription, so you don't need your own GPU infrastructure.
Pro costs $5/month (billed yearly) for 300 credits, and Premium costs $20/month (billed yearly) for 1,200 credits — roughly 60 to 240 standard videos. One-time credit packs are also available, from $1.50 for 30 credits up to $100 for 3,900 credits.
Yes, starting with Wan 2.5. The model generates synchronized dialogue, sound effects, and lip movement in the same pass as the video, rather than requiring a separate audio generation step.
As of July 2026, Wan 2.7 is the latest release, following Wan 2.6 (December 2025, multi-shot storytelling and consistent character identity) and Wan 2.5 (September 2025, native audio generation).
Yes. Alibaba Cloud's Model Studio (DashScope) offers official API access to Wan models, priced around $0.10-$0.15 per 5-second 720p clip, separate from the wan.video consumer subscription.