Back to all posts

OpenAI Sora 2 Upgrade Rumored With GPT 5 Release

2025-07-25Alex Hughes3 minutes read
Ai Video
Openai
Sora

The buzz around OpenAI's upcoming GPT-5 is undeniable, but a new rumor suggests it might not be coming alone. Speculation is mounting that a major upgrade to Sora, OpenAI's text-to-video model, could be part of the package, and it might be the most exciting update yet.

OpenAI

The Sora 2 Leak What The Rumors Say

The primary clue comes from Tibor Blaho, an AI commentator known for his industry insights. He shared a compelling screenshot on X which appeared to show host names on OpenAI.org containing files named "Sora 2." This has fueled widespread belief that a second version of the powerful video generator is in the works.

Sora made a huge impact when it was first revealed, representing a massive step forward for AI video generation. However, the AI landscape moves incredibly fast. Since its initial launch, Sora has seen little public change, while competitors have surged ahead. Models like Google’s Veo 3, as well as impressive tools from Kling, Runway, and Moonvalley, have pushed the boundaries of what's possible. It makes perfect sense for OpenAI to bundle a Sora update with GPT-5 to avoid falling behind in this critical area of AI development.

What Could a Sora 2 Upgrade Include

While nothing is confirmed, we can speculate on the features a new version of Sora would need to compete. Here are the most likely improvements we could see.

Better Physics and Movement

Sora video of two people playing tennis

A common criticism of the current Sora is its poor grasp of physics and human movement. Videos often show bizarre artifacts, like people melting into their surroundings or limbs moving in impossible ways. This happens because AI models are trained on video data without a fundamental understanding of physical laws. For example, recreating a surfer is difficult if the model doesn't understand how a person, a board, and water interact. Fixing this "physics problem" would be a major breakthrough for Sora 2.

Adding Sound to the Picture

phone with Chatgpt logo

One of the key advantages that models like Veo 3 have is the ability to generate audio along with video. Early AI video tools were silent, but newer competitors are integrating sound for a more complete and immersive final product. For Sora 2 to be truly competitive, OpenAI will likely need to add an audio generation feature. The company has experimented with AI audio in the past, so this seems like a natural next step.

Generating Longer Video Clips

Currently, most AI video generators can only create clips that are four to eight seconds long. If OpenAI could enable Sora 2 to produce significantly longer videos, it would be a major differentiator. This would be a computationally expensive feature, so it would almost certainly be reserved for premium subscribers, but it would be a huge leap forward for AI filmmaking.

A Boost in Video Quality

Moonvalley AI

A recent trend in AI video is a significant increase in footage quality. Competitors like Moonvalley are making a name for themselves by producing exceptionally high-resolution video, reportedly by training their models exclusively on high-quality footage. While OpenAI may not take such a specialized approach, we can expect the general trend of technological improvement to result in Sora 2 producing sharper, more detailed, and more realistic video.

Read Original Post
ImaginePro newsletter

Subscribe to our newsletter!

Subscribe to our newsletter to get the latest news and designs.