OpenAI's Sora 2: Ready to Beat Google's Veo 3?

We may soon see a new AI that not only creates amazing videos from text but also adds realistic sound to them—speech, ambient noise, and even lip-syncing. That is exactly what is now being discussed in connection with OpenAI's upcoming Sora 2. According to information from the company's servers discovered by analysts, a major update is in the works that is expected to bring sound to text-generated videos. All of this is intended to help Sora compete with Google's advanced Veo 3 model. Let's take a closer look, step by step, based on the details available from TechRadar and other sources.

What Do We Know About Sora 2 from the Available Information?

According to reports from TechRadar and information found on OpenAI's servers, Sora 2 has appeared in internal references, suggesting that its release is imminent, although there has been no official announcement yet. The first version of Sora impressed with its ability to create videos of high visual quality featuring realistic physics and longer sequences—specifically, clips longer than 20 seconds. This is a major advantage over competitors, which can often handle only shorter segments. But Sora 1 was "silent"—no sound, no dialogue, just visual content. Sora 2 is now expected to address this shortcoming. Sources suggest the integration of sound, including speech, ambient noise, and lip-syncing, which could create a fully-fledged cinematic experience. This update could be linked to the GPT-5 model, which would help improve the overall quality and integration.

Comparison with Google's Veo 3: Where Is Sora Falling Behind?

Veo 3 is already demonstrating what advanced sound integration looks like. This model generates videos with realistic sound effects—from coffee splashing to café noise—and adds dialogue with precise lip-syncing. Videos from Veo 3 are limited to eight seconds, but their quality reaches 4K with a cinematic feel. According to TechRadar, Sora 2 must not only add sound but do so to a high standard: with natural intonation, harmony between visuals and voice, and the right atmosphere. If OpenAI maintains the longer video length and adds these elements, Sora 2 could gain an advantage in creative and professional projects such as education or research. Veo 3, on the other hand, is integrated into the Google ecosystem, available free to students through Gemini AI Pro and to others through Vertex AI on Google Cloud.

Expected Sora 2 Features

The key improvement in Sora 2 should be full sound integration, meaning not only speech but also environmental sounds and lip-syncing. This would make it possible to create videos that are not only visually impressive but also aurally realistic. Sources such as TechRadar emphasize that Sora 2 must "dazzle" viewers to surpass Veo 3, which already offers a synchronized audio track. Another factor is price: Veo 3 is part of the Gemini Advanced subscription costing $250 per month, while Sora 2 could be integrated into the paid ChatGPT Plus or Pro plans, which could make it more affordable. If OpenAI provides easier access, it could quickly attract the general public.

Adding voices and realistic videos also brings risks. Both models—Sora and Veo 3—have restrictions: they prohibit generating content featuring real people, violence, or copyright infringement. Nevertheless, adding voices increases the risk of misinformation because the videos could look and sound too convincing. Companies such as OpenAI and Google are already addressing these issues, but as AI-generated videos become increasingly sophisticated, ethical questions are becoming even more pressing. For example, related reports discuss how Sam Altman is reassuring Hollywood filmmakers about Sora's capabilities.

This competition between OpenAI and Google could set new standards in generative video, where it is no longer just about pixels but the overall cinematic effect. If Sora 2 lives up to expectations, exciting advances in AI technology await us. Stay tuned for the official announcement!