This page is only in Korean for now. Your browser can translate it, or go to the English home.
Technologies
글·그림·참조 영상으로 새로운 영상을 만드는 AI예요.
인물 일관성, 소리와 대사, 편집 제어처럼 실제 제작에 필요한 변화를 봐요.
Up to 40 recent matching stories, newest first. Newly collected stories appear here automatically.
Gathering stories…
Search earlier stories →Names are matched in the titles and summaries of public stories collected by LastAI. Some stories may be missed or refer to a namesake. Coverage is not exhaustive. Profiles updated .
BytePlus's AI video model Seedance 2.5 can now create 4K video inside Picsart. It can produce a single scene up to 30 seconds long in 4K.
Users who claim to be connected to Anthropic's unreleased AI model 'Claude Page 5.5' have posted animations and 3D scenes made only from code on X one after another. They say the AI-written code produced the scenes and music, without any image or video generation AI. Anthropic has not confirmed that this model exists.
Google has released Gemini Omni 1.2 Flash, a video generation model. It can create videos up to 15 seconds long, and its price has dropped by 30%. The model also introduces a feature for editing videos through conversation. In the app, only Ultra subscribers can use it.
Google has officially released Gemini Omni 1.2 Flash, its AI video generation model. The maximum length of a video made at once has increased from 10 seconds to 15 seconds, and the price at 720p has dropped 30% to $0.07 per second. Chat-based video editing has also been strengthened.
In its Marketing Guide corner, Review Times shows how to use Google FLOW (an AI image and video creation tool) for free on a Galaxy phone. It also covers making 4 to 5 videos a day with free image generation and veo 3.1 lite (an AI video generation model).
This video from the Matt Wolfe channel gathers the week's AI news in one place. It briefly introduces newly released features, including image generation tools, video generation, and phone call functions.
Google has started offering Gemini Omni 1.2 Flash, a video generation AI, to the public. It can create videos up to 15 seconds long in one go, and users can refine the video over several rounds by giving spoken instructions. The price for 720p video is 30% lower than the previous version.
Google has released Gemini Omni 1.2 Flash, a new model that creates AI videos up to 15 seconds long and lets users edit them through conversation. Anyone with a Google account can make up to three videos per month for free, and the cost of generating videos is 30% lower than before.
Google has released Gemini Omni 1.2 Flash, a video generation AI. It can now create videos up to 15 seconds long in one go, and its usage cost is 30% lower than the previous version. Google says it has improved the tool for commercial videos such as advertisements.
Google has released Gemini Omni Flash 1.2, a new version of its video model. It creates clips with built-in sound and can extend an existing video. Higgsfield has made the model available on its own platform.
Google has officially released Gemini Omni 1.2 Flash, a video generation and editing model. The default clip length has increased to 15 seconds, and a conversational editing feature lets users refine videos over multiple rounds. The price for 720p has dropped by 30%.
Google DeepMind has released Gemini Omni 1.2 Flash, a video generation AI. It can now create videos up to 15 seconds long in one go, and usage costs are 30% lower than the previous version. A new feature also lets users revise videos step by step in a back-and-forth conversation.
A ranking and comparison of AI video generators for 2026. Google Veo 3.1, Kling 4.0, and Seedance 2.5 are presented as strong starting points for general video production. Runway is described as a good fit for projects that need more detailed editing control. The original text cuts off partway, so the article's information on HeyGen and Synthesia cannot be confirmed.
Using the Neuron Kernel Interface, a Reactor–AWS collaboration tackled the dynamic shapes, memory access patterns, and cache management that make real-time autoregressive diffusion hard—building techniques that generalize across models.
