Sora 2 could be the biggest update to OpenAI’s text-to-video models: the company is working on releasing a new tool that can compete with Google’s powerful Veo 3. This is evidenced by mentions of Sora 2 seen on OpenAI’s servers, although there has been no official announcement yet.
Sora 2 could be the biggest update to OpenAI’s text-to-video models: the company is working on releasing a new tool that can compete with Google’s powerful Veo 3. This is evidenced by mentions of Sora 2 seen on OpenAI’s servers, although there has been no official announcement yet.
According to TechRadar, the key challenge for the second generation of Sora will be not only visual quality, but also full integration of sound: speech, environmental noise, and lip-sync. These are the capabilities that Google previously demonstrated in its Veo 3: the model generates videos with realistic sound effects, dialogue, and atmosphere.
The first version of Sora was impressive with its images, but remained a «silent movie». Veo 3, on the other hand, already shows a synchronized audio track: from the sound of coffee splashing to the noise of a coffee shop — all this creates a full-fledged cinematic experience. At the same time, Google videos are limited to eight seconds in length, while Sora is capable of generating clips lasting more than 20 seconds.
To outperform its competitor, Sora 2 must not just «speak,» but do it well: with realistic intonation, harmony between image and voice, accurate lip-sync, and the right atmosphere. If OpenAI can bring the quality to at least the level of Veo 3, while offering longer videos, this could give the model a serious advantage in creative and professional projects.
Price also plays a role. Veo 3 is currently only available as part of Google’s Gemini Advanced subscription, which costs $250 per month. OpenAI will likely integrate Sora 2 into its paid ChatGPT Plus or Pro plans. If access proves cheaper or easier, the audience will grow rapidly.
As AI video becomes more sophisticated, ethical questions arise, particularly around generating believable voices and images. Both companies have already restricted videos from featuring real people, violence, or copyrighted content. But adding voiceovers only increases the risk of misinformation.
Despite these challenges, it seems that the release of Sora 2 is just around the corner. And this battle of the giants could set new standards in the field of generative video, where the competition is no longer about pixels, but about cinematic effect.
As a reminder, we also published an article about how Microsoft Bing is launching Bing Video Creator in its app, which uses OpenAI’s Sora model. Users will be able to create videos from text prompts.
