If you build with AI video, you already know the cycle.
A new model drops.
Quality jumps.
Then the real question shows up in the same meeting every time.
“Cool. How much does it cost per second?”
Today, we have a clear answer and a big update.
We are bringing Alibaba WAN 2.7 to Picsart AI Hub, and we are launching it with special pricing at 10 credits per second.
For comparison, many other video models in AI Hub are priced at 50 credits per second.
Yes, that is the kind of difference your finance team will actually smile at.
Wan is an open video generation model family released by Alibaba Cloud, built for text to video and image to video workflows.
And inside Picsart, WAN 2.7 is not just “available.” It is becoming the default experience in our video generation products.
What is WAN 2.7, and why teams are paying attention
WAN 2.7 is the latest iteration of the WAN video generation model. In Picsart, it is positioned as a controllable model that helps teams steer results more precisely instead of gambling on endless reruns.
Here is the practical difference.
A lot of video generation tools are great at making something interesting once.
WAN 2.7 is built for getting something usable repeatedly.
In Picsart, WAN 2.7 supports:
- Multi reference inputs, including up to 9 reference images or 5 reference videos in a single generation
- First and last frame control, so you can define how a clip starts and ends
- Instruction-based editing, so you can change what is “almost right” without starting over
- Video recreation, which keeps motion and pacing while generating variations
- Subject plus voice locking, for more consistent characters across outputs
That is a big deal if you are building ads, product clips, character based content, or any workflow where consistency matters more than novelty.
Picsart also notes that WAN 2.7 is available to a very large creator base on the platform, and every output can move directly into editing and publishing flows.
The official WAN positioning, in plain language
On the WAN side, the theme is “control.” Less guessing, more directing.
Here are the capabilities WAN highlights, rewritten in everyday terms:
- Edit video with instructions
Rework scenes with text prompts, and use multiple images for guidance, so the model follows your intent more closely. - Transfer motion and camera style
Replicate the dynamics of a reference, from character motion to cinematic camera movement. - Extend time and continue scenes
Keep a clip going with frame guided continuation and outpainting style extension. - Dial in portraits and faces
Control details like facial structure and subtle features so characters look intentional. - Render text and graphics
Better support for longer text in multiple languages, plus structured elements like charts and formulas. - Control color distribution
More predictable color placement so your output matches a specific look. - Multi image editing and fusion
Combine multiple images into one coherent creative direction. - Sequential storytelling
Generate a consistent sequence with the same subject and style. - Interactive edits
Use box selection style controls to steer changes in a specific region.
If you have ever tried to generate a “consistent character in a consistent scene” and ended up with “their cousin in a different decade,” you get why this matters.
The pricing update, and why we are doing it
Here is the headline again, because it deserves it.
WAN 2.7 on AI Hub is 10 credits per second.
Many other video models on AI Hub are 50 credits per second.
This launch rate is exclusive to AI Hub.
If you are new to Picsart credits, they are simply the usage unit for generative tools. Think of them as tokens you spend to generate results.
We are doing this for the same reason we built AI Hub in the first place.
Teams want to use the best models. They also want to stay sane.
AI Hub is designed as a single gateway to a large catalog of creative AI models, so you can keep one integration and still evolve your stack as models change.
Why AI Hub makes this model more valuable
A new model is great.
A new model that does not force a new integration is better.
AI Hub is built so teams can:
- Use one API surface for image, video, text, and audio models
- Compare quality, speed, and cost side by side using consistent patterns
- Switch models by changing the model identifier, not by rewriting workflows
- Pin versions so production outputs stay stable even as models evolve
- Start with evaluation credits, since eligible accounts can receive 200 complimentary credits for testing
So you can start with WAN 2.7 as your default baseline, and still keep the freedom to switch later if a different model wins for a specific use case.
That is the core promise.
Build once. Stay flexible.
Enterprise readiness, without the extra drama
A lot of teams love experimenting with models.
Enterprise teams need the experiment to survive procurement, security review, and scale testing.
AI Hub is built for enterprise rollouts, and Picsart provides compliance signals through its Trust Center, including SOC 2 Type II, SOC 3, ISO 27001, and GDPR support.
That matters if you are moving beyond prototypes into production workloads.
What happens next
This update is part of a bigger commitment.
At Picsart, we are doing the work to bring the latest models into one place, keep the integration consistent, and make experimentation more affordable so teams can actually use what is new.
WAN 2.7 is a strong default for AI video today.
If you want to try it, the easiest path is:
- Start with AI Hub access
- Run a few test generations
- Keep WAN 2.7 as default, or switch models later without changing your pipeline
