Skip to main content

Digital Trends may earn a commission when you buy through links on our site. Why trust us?

This new AI video generator just completely outdid Sora

An AI-generated image of a woman with blonde hair.
An AI-generated image created by KlingAI. Kuaishou Technology

Sora’s lead in the generative video space is already in jeopardy, and Meta’s highly anticipated AI hasn’t even been released yet. On Wednesday, Chinese tech titan Kuaishou Technology announced that its Kling AI video-generation system is now available to users around the globe.

Kling had officially launched in China last month, but was only available for users in the country and required a Chinese phone number to access. With the global rollout, users simply need to provide an email address — and a fair amount of patience.

Recommended Videos

Here we go! Kling AI is now available globally.

People have been already creating wild AI videos.

10 examples:

1. Grannypic.twitter.com/9n6MB4jOYK

— Min Choi (@minchoi) July 24, 2024

The model is capable of generating high-definition video clips up to two minutes in length — double what Sora will reportedly be able to produce — at 30 frames per second (fps) and 720p resolution. Kling offers a 2,000-character context window and can receive both text and image inputs to create a quick 5-second clip. Users can then consecutively extend their videos by 4.5 seconds at a time, according to the company’s website. The Kuaishou team wrote on X that users will receive 66 credits per day. Each video generation will cost 10 credits. The team is also reportedly developing a subscription model.

Despite its impressive generative capabilities, Kling is limited in what it can produce. Specifically, it’s unable to generate videos of a politically sensitive nature. According to a TechCrunch report Thursday, prompts like “Democracy in China,” “Chinese President Xi Jinping walking down the street” and “Tienanmen Square protests” return nonspecific error messages. However, users are still able to upload and animate images of the same subjects without issue, so long as the user didn’t name the subject specifically in their prompt.

The Financial Times noted earlier this month that China’s primary internet regulator, the Cyberspace Administration of China (CAC), plans to begin purity testing of AI systems developed in-country to ensure they “embody core socialist values” on a variety of sensitive topics. Those include criticism of the Communist Party and President Xi Jinping himself.

Kling is also severely limited, at least currently, in how quickly it can produce content. A five-second clip can take as much as 15 minutes or more to generate. Even getting the system to send an email verification code took inordinately long. It’s not clear, however, if that lag is due to underlying infrastructure issues or simply enormous user demand as folks rush to try out the new AI.

Kling isn’t the only AI to rival Sora. Runway released its Gen-3 video AI, which is capable of creating clips up to 10 seconds in length, earlier this month. That followed Luma Labs’ debut of Dream Machine in June, which saw similar performance issues when users overloaded the platform at launch.

Andrew Tarantola
Former Digital Trends Contributor
Andrew Tarantola is a journalist with more than a decade reporting on emerging technologies ranging from robotics and machine…
I tested the future of AI image generation. It’s astoundingly fast.
Imagery generated by HART.

One of the core problems with AI is the notoriously high power and computing demand, especially for tasks such as media generation. On mobile phones, when it comes to running natively, only a handful of pricey devices with powerful silicon can run the feature suite. Even when implemented at scale on cloud, it’s a pricey affair.
Nvidia may have quietly addressed that challenge in partnership with the folks over at the Massachusetts Institute of Technology and Tsinghua University. The team created a hybrid AI image generation tool called HART (hybrid autoregressive transformer) that essentially combines two of the most widely used AI image creation techniques. The result is a blazing fast tool with dramatically lower compute requirement.
Just to give you an idea of just how fast it is, I asked it to create an image of a parrot playing a bass guitar. It returned with the following picture in just about a second. I could barely even follow the progress bar. When I pushed the same prompt before Google’s Imagen 3 model in Gemini, it took roughly 9-10 seconds on a 200 Mbps internet connection.

A massive breakthrough
When AI images first started making waves, the diffusion technique was behind it all, powering products such as OpenAI’s Dall-E image generator, Google’s Imagen, and Stable Diffusion. This method can produce images with an extremely high level of detail. However, it is a multi-step approach to creating AI images, and as a result, it is slow and computationally expensive.
The second approach that has recently gained popularity is auto-regressive models, which essentially work in the same fashion as chatbots and generate images using a pixel prediction technique. It is faster, but also a more error-prone method of creating images using AI.
On-device demo for HART: Efficient Visual Generation with Hybrid Autoregressive Transformer
The team at MIT fused both methods into a single package called HART. It relies on an autoregression model to predict compressed image assets as a discrete token, while a small diffusion model handles the rest to compensate for the quality loss. The overall approach reduces the number of steps involved from over two dozen to eight steps.
The experts behind HART claim that it can “generate images that match or exceed the quality of state-of-the-art diffusion models, but do so about nine times faster.” HART combines an autoregressive model with a 700 million parameter range and a small diffusion model that can handle 37 million parameters.

Read more
OpenAI releases a new AI model, but it’s eye-wateringly expensive
OpenAI's new typeface OpenAI Sans

OpenAI has released its latest model, o1-pro, an updated version of its reasoning model o1 -- but it's not going to come cheap.

"It uses more compute than o1 to provide consistently better responses," OpenAI said in its announcement. The company went on to say it offered new features including: "Supports vision, function calling, Structured Outputs, and works with the Responses and Batch APIs. "

Read more
Razer’s new AI-powered bug detector could help games release faster
Razer Blade 16 2025

Razer has announced a new AI-powered game development tool called AI QA Copilot, which aims to help QA testers find and report bugs faster.

A faster QA cycle could help developers get games out of the door in less time too so if it works well, it would be good for everyone. Bugs appear everywhere in games during development -- they're impossible to avoid. QA (Quality Assurance) teams are dedicated to finding problems and creating detailed reports to help get things fixed faster.

Read more