OpenAI has just released an advanced version of its groundbreaking Sora AI model, named Sora Turbo, marking a significant leap in text-to-video technology. Initially introduced earlier this year, Sora was designed to simulate and understand the physical world, providing a foundation for AI that can interact with reality. Now, with Sora Turbo, OpenAI is pushing the envelope further by offering enhanced speed and a suite of new features.
Sora Turbo is accessible through a dedicated website, Sora.com, exclusively for ChatGPT Plus and Pro subscribers. Users can now generate videos with resolutions up to 1080p, with a maximum duration of 20 seconds, in various aspect ratios including widescreen, vertical, or square. This model allows users to not only create new content from text but also to extend, remix, and blend their own assets into their videos.
New interfaces have been developed to simplify the process of prompting Sora with text, images, or videos. A notable addition is the storyboard tool, which enables users to specify detailed inputs for each frame, enhancing control over the video narrative. Additionally, there are now Featured and Recent feeds where the community’s creations are showcased, fostering a collaborative environment.
For those on a Plus plan, Sora is included without extra cost, allowing for the generation of up to 50 videos at 480p resolution monthly, or fewer at 720p. Pro subscribers get amplified benefits, including 10 times more usage, higher resolutions, and longer video durations. OpenAI is also working on customized pricing for different user types, with plans to introduce this early next year.
Despite these advancements, Sora Turbo still has its limitations, particularly in simulating realistic physics and handling complex actions over extended periods. OpenAI acknowledges these challenges and is committed to making the technology more accessible and affordable.
The release of Sora at this stage is strategic, aiming to give society time to understand and adapt to this powerful tool responsibly. To ensure transparency and authenticity, all Sora-generated videos include C2PA metadata for origin verification. While not foolproof, additional safeguards like visible watermarks and an internal search tool have been implemented to help identify AI-generated content.
Safety is paramount, with restrictions on creating and uploading content like child sexual abuse material (CSAM) and sexual deepfakes. Initially, uploads involving people are limited, with plans to expand this feature as better deepfake mitigations are developed. More details on these safety measures can be found in the system card and through OpenAI’s red teaming efforts.
Sora is currently not available for users under 18, nor for those in the UK, Switzerland, or the European Economic Area. However, OpenAI is actively working on expanding access. Looking ahead, partnerships with NGOs and research organizations are on the horizon to enhance the digital content provenance ecosystem, aiming to build trust and authenticity in digital media.
With Sora Turbo, OpenAI invites the world to explore new creative avenues and storytelling possibilities, all while navigating the ethical landscape of AI in video production.
Your First 10 AI Skills
10 practical AI skills, copy-paste prompts and a 7-day plan to start using AI with confidence.
Download the guide →