On December 10, OpenAI made a significant announcement regarding the accessibility of its video generation model, Sora, to paid ChatGPT users in the United States and select regions, excluding markets like Europe and ChinaThis latest iteration, Sora Turbo, boasts the capability of producing videos that can last up to 20 secondsThe pricing for subscriptions featuring this new model stands at $20 per month for ChatGPT Plus and $200 for ChatGPT Pro, allowing users to create up to 50 and 500 videos respectivelyThis launch marks a major milestone, showing that OpenAI's video generation initiative has finally transitioned from mere showcase to user engagement following a 10-month wait.

Initially revealed in February of this year, Sora garnered significant attention, primarily among professional users such as artists, animators, and filmmakersThe recent broader rollout led to a surge in registration on the Sora website, resulting in technical difficulties for many users trying to access the platform

Rohan Sahai, the product lead for Sora, commented on this situation, acknowledging the overwhelming demand and indicating that registration functionalities were temporarily disabled, although he expressed optimism about fully integrating these capabilities soon.

"Generational Gaps Not Apparent"

Since Sora’s debut, dozens of companies globally have entered the AI video generation space, including notable competitors such as Kuaishou's Keling AI, Runway, and Vidu by ShengShu TechnologyJust recently, Tencent also joined the landscape with its open-source model, while ByteDance initiated beta testing for its video generation model “Doubao.” The market is heating up, and notable advancements have been achieved in areas such as generation quality, clarity, realism, and duration across various AI video generation products.

A Chinese proverb states, "Only when you pull out a horse can you see if it’s a donkey or a horse." Now, nearly ten months after Sora's initial introduction, the question arises: Is Sora still the undisputed market leader?

According to Shen Yang, a dual-hired professor at Tsinghua University’s School of Journalism and School of Artificial Intelligence, “Currently, Sora remains a leading player in the field of AI video models, but it hasn't created a generational gap with its competitors.” By examining leading models like Keling AI and Runway, he notes that after several months of iterations and enhancements in user experience, these competitors have developed a range of “moat” functionalities

• Rate Cuts Mark Major Shift by the Fed
• US Stocks Face Headwinds in H2 Next Year
• Bank of Japan Holds Steady on Interest Rate Hikes
• Global Currency War Escalates as RMB Surges 500 Points
• Why the Russian Ruble Sharply Devalued

The advantages displayed by Sora, arriving late to the market, are not significantly pronounced, especially regarding its capabilities in generating “videos from images,” where it still lags behind.

Shen mentions that Sora’s primary strength lies in its association with the vast user base of ChatGPT, which effectively channels users towards Sora, resulting in a higher quality of visuals due to the enhanced enabling features from ChatGPT promptsHowever, the user counts for Runway and Keling AI remain significantly lower than that of OpenAI’s suite of productsPost-launch, Sora's servers experienced overwhelming demand, showcasing the draw of OpenAI products in general, suggesting that Sora’s user base could surpass that of competitors like Runway and Keling in a matter of months.

Reports from various US media outlets reveal that OpenAI's CEO, Sam Altman, disclosed in December that ChatGPT now boasts over 300 million weekly active users, processing over a billion messages from users daily

While Runway has not released any user statistics, traffic analysis from SimilarWeb noted 7.2 million visits to the Runway site in OctoberKuaishou announced in its Q3 report for 2024 that Keling AI has surpassed five million users since its June launch, with over 1.5 million active users in September alone, generating over 51 million videos and more than 150 million images.

“OpenAI has finally opened Sora after a 10-month wait,” stated Lei Tao, CEO of innovative video model firm New Yi TechnologyHe noted that this move reflects OpenAI’s urgent need to cultivate a broader array of successful core products to expand its commercial avenues and enhance revenue and visibility in the capital markets.

Fan Jiarui, head of investment and financing at ShengShu Technology, which has also been in the AI video generation space for a while now, expressed strong confidence in their own model, Vidu

He asserted that while Sora initially led the industry, market developments over the past ten months have resulted in Vidu standing on equal footing, each with its distinct featuresThe Vidu platform has gained rapid popularity across global social networks with creative offerings like “Embrace Your Past Self” and “One-Click AI Transformation,” simultaneously streamlining content creation processes for professional entrepreneurs and greatly improving video production efficiency.

Early Signs of Application Value?

During the past 10 months, has the rapid advancement of AI video generation technology shown its impact on film production, advertising, and content creation?

Shen Yang believes that the release of video models such as Sora and Keling AI marks a deeper integration of AI video generation technology into the realms of film production, advertising, and content creation

The developments of the past year indicate that AI has transitioned from a theoretical or experimental phase into one that significantly influences every step of the creative process.

In film production, traditional workflows are being transformedMany filmmakers are now leveraging AI as an effective tool for visualizing scriptsFor instance, the world’s first 90-minute film "Our Terminator 2 Remake," created entirely by AI, has already been released in the U.SNumerous cases of AI-generated advertisements, travel videos, public service announcements, and corporate promo videos aboundShen's team has completed several AIGC (AI-generated content) projects this year including China’s first AI micro-drama "Chinese Myths," a tourism promotion video for Jiangning, a public service short for Shanghai’s anti-drug commission titled "A Century of Anti-Drug, Success Belongs to the Committed," and a sci-fi short film "Prometheus," which won the Best AI Film Editing Award at the 2024 International Student Media Arts Festival (ISMA), among over 300 creative content pieces aimed at serving society.

Fan Jiarui shared that Vidu is currently focused on enhancing the consistency and generation speed of objects in videos to grant creators more freedom

For practical applications, Vidu has opened its API services and partnered with key clients across sectors like entertainment, advertising, and film animation, establishing hands-on collaborations.

Lei Tao mentioned that multiple video AI generation firms are actively engaged in the domestic market, with some capturing significant shares in the consumer sectorMeanwhile, New Yi Technology's strategy is more concentrated on the business-to-business (B2B) market, where they consider practical applications of AI technology across client operationsWith ongoing technological advancements, the video generation sector will increasingly prioritize content authenticity, coherence, and personalizationThe spectrum of application scenarios will continue to expand, benefitting fields ranging from advertisement production and film creation to corporate promotionParticipating companies will also deploy more API interfaces and customized services to leverage video generation technologies in diverse scenarios, thus enabling revenue growth.

However, Lei pointed out from a user experience perspective that compared to text-based or image generation, the upgrading of AI video generation models this year has not been immense, and the market participation remains somewhat limited

This limitation is largely attributed to the current nascent stage of AI video generation, where both technology and application are still maturingThe complexity involved in video generation surpasses that of text and images, demanding higher investmentsVideo projects must handle vast amounts of data, including visuals, audio, and movements, while maintaining the coherence and authenticity of content and subjectsAdditionally, the burgeoning demand for high-quality, personalized video content among users poses further challengesAlthough the potential of the video generation market is immense, the R&D costs and high technical barriers significantly constrain the investment capabilities of companies.

Shen Yang concluded by emphasizing that the past year's applications of AI video generation technology have unveiled substantial value, particularly in enhancing creative efficiency and flexibility