Google Launches Gemini Omni 1.2 Flash: Affordable AI Video Generation Up To 15 Seconds Long

Google has unveiled its Gemini Omni 1.2 Flash video-generation model, making it widely accessible for users to create short clips up to 15 seconds long. This development represents a notable advancement in the field of artificial intelligence, offering both enhanced capabilities and more economical pricing compared to previous iterations.

The updated model boasts a reduced price point, with the cost per second dropping to $0.07 for 720p video, marking a 30% decrease from earlier rates. This makes the service more affordable for a broader range of users, including developers and businesses looking to integrate dynamic video content into their offerings.

Gemini Omni 1.2 Flash also brings improvements in visual consistency and fluidity of dialogue, allowing for smoother transitions and more natural-looking animations. Additionally, the model can now handle complex camera movements with fewer abrupt changes, enhancing the overall quality of the generated videos. These enhancements make the new model particularly suitable for applications where smooth, lifelike video is crucial.

One of the standout features of Gemini Omni 1.2 Flash is its ability to generate videos based on various types of input, such as character images, product photos, short motion clips, and even audio cues. This versatility opens up numerous possibilities for creative expression and practical use cases, from marketing materials to educational content.

Moreover, the model’s reference adherence has seen a significant improvement, meaning that the generated videos closely match the intended visuals and sounds provided by the user. This increased accuracy is especially beneficial for scenarios where precise representation is essential.

Another notable feature is the inclusion of an imperceptible SynthID watermark in all generated videos, ensuring that the source of the content can be traced back to Google’s technology. This adds an extra layer of authenticity and verification to the videos created using Gemini Omni 1.2 Flash.

However, there are some aspects of the new model that remain unconfirmed. For instance, while it is reported that the model is now generally available for use in Google’s products and developer platforms, this has yet to be officially verified. Similarly, the exact range of clip lengths and the availability of multi-turn editing features are still under investigation.

In summary, Google’s Gemini Omni 1.2 Flash offers a compelling combination of affordability, quality, and versatility, setting a new standard for AI-generated video content.

What we know

• - Google's Gemini Omni 1.2 Flash video-generation model is now generally available. ¹²³⁴
• - The model creates clips up to 15 seconds long. ¹²³⁴
• - Pricing for the model has been reduced by 30 percent. ¹²³⁴
• - The new pricing is $0.07 per second for 720p video. ¹²³⁴
• - Reference adherence is significantly better with the new model. ¹²³⁴
• - Clips can now range from 3 to 15 seconds in length. ¹
• - The model can use character images, product photos, short motion clips, and audio cues as inputs. ¹
• - Visual consistency and fluidity of dialogue have improved. ¹
• - The model can handle complex camera movements with fewer jump cuts. ¹
• - Users can now perform multi-turn editing across full 15-second clips without needing to regenerate the entire clip. ²
• - Gemini Omni 1.2 Flash is available on Google Cloud and through the Gemini API. ²
• - Gemini Omni 1.2 Flash is hosted by Google only, with no open weights or local deployment options. ²
• - Generated videos carry an imperceptible SynthID watermark. ¹

What remains unclear

• - Whether Gemini Omni 1.2 Flash is now generally available for use in Google's products and developer platforms. ³⁴