New models and features in AI image, video, voice and music generation, from Midjourney, Sora, Veo, Suno and more.
Alibaba researchers just unveiled EMO, an expressive audio-driven portrait-video generation framework. Give it a single still image and an audio clip—talking or singing—and it animates the portrait with realistic facial expressions and head poses for any duration. It even keeps up with fast-paced rap and works across multiple languages, paintings, and 3D models.
ElevenLabs just partnered with Perplexity to launch Discover Daily, a short-form daily podcast covering breakthroughs in innovation, science, and culture. The show merges Perplexity's real-time search engine with ElevenLabs' synthetic voices to deliver automated audio briefings on the go, available now across major podcast platforms.
Suno just rolled out early access to its V3 Alpha model for Pro and Premier subscribers. The update bumps maximum clip lengths to two minutes, adds dedicated instrumental support, improves audio quality, and lets creators continue tracks from any point. Paying accounts also received 300 free credits to test the model.
Stability AI just announced Stable Diffusion 3 in early preview, calling it its most capable text-to-image model yet. The suite ranges from 800M to 8B parameters and combines diffusion transformer architecture with flow matching to solve messy spelling and multi-subject prompts. A waitlist is now open ahead of an open release.
Will Smith just dropped a hilarious Instagram clip parodying the infamous 2023 AI video of him devouring pasta. Coming right after OpenAI debuted Sora, Smith's real-life recreation highlights how quickly AI video generation evolved from a surreal nightmare meme to uncanny realism.
Microsoft just confirmed that OpenAI's Sora video generator will eventually make its way to Copilot. Advertising and web services CEO Mikhail Parakhin replied on X that integration is planned, though it will take time. There is no official timeline yet, but bringing text-to-video directly into Copilot promises to significantly expand its multimodal creative toolkit.
Stability AI just launched Stable Cascade in research preview, a text-to-image model built on a three-stage compression architecture that cuts fine-tuning costs by 16x compared to Stable Diffusion. Training, ControlNet, and inference scripts are live on GitHub, though it remains under a non-commercial research license.
Midjourney just launched alpha testing for its dedicated image creation website, opening access to anyone who has generated more than 1,000 images. The team is focusing primarily on desktop before rolling out mobile support, with plans to expand access more broadly as the platform matures.
Stability AI just launched SVD 1.1, an upgraded image-to-video diffusion model tuned for better motion and visual consistency. Based on SVD-XT, it generates four-second, 25-frame clips at 1024x576 resolution from a single still frame. The weights are free for research on Hugging Face, while commercial use requires a paid membership or API access.
Google just made its experimental music generator MusicFX generally available in AI Test Kitchen, alongside lyric-writing helper TextFX. The tool turns text prompts into tracks up to 70 seconds long or seamless loops, tagged with DeepMind's SynthID watermark. Why it matters: Google is testing the waters of creative audio while carefully blocking vocal generation and artist name prompts to avoid copyright backlash.
Creative Works London just revealed how they created the new music video for Guns N' Roses' track 'The General' using a hybrid generative workflow. The team built base scenes in Unreal Engine and fed them into Stable Diffusion to dream over each frame, creating uncanny, shifting imagery. It offers a practical blueprint for artists treating AI as a creative amplifier rather than a replacement.
Midjourney just launched Niji V6, a new version of its latest model fine-tuned specifically for anime-style artwork. It brings vastly improved comprehension of intricate prompts, direct in-image text generation, and a Style Raw option, with inpainting and zoom tools slated to roll out in February.
Runway just rolled out Multi Motion Brush to all Gen-2 users, letting creators control independent movements across different parts of a static image. Users can brush several distinct zones and use sliders to tweak horizontal, vertical, and proximity motion for each layer. It gives video creators precise directional control rather than leaving the whole scene to AI guesswork.
Midjourney just shared a major roadmap update in its office hours, confirming that inpainting, pan, and zoom are returning to v6 as early as next week. Style consistency will arrive before character consistency, and the team is prioritizing video generation over 3D. Control and consistency are finally catching up to the raw fidelity of v6.
Amazon just rolled out an AI Art feature to select Fire TV devices in the US, turning the big screen into a customizable canvas. Powered by Amazon's fine-tuned Titan Image Generator, users can simply ask Alexa to generate up to four tailored images in styles like watercolor or pixel art, and set their favorite directly as the ambient background.
Midjourney CEO David Holz just announced that the team will start training video generation models in January. On top of that, Midjourney v6 is getting an update next week to improve text rendering and prompt following, while gradually restoring v5 staples like inpainting. The reigning king of image generation is finally moving into motion.
Pika Labs just made its text-to-video platform, Pika 1.0, available to everyone on the web. The tool lets users create 3D animations and cinematic clips from prompts, complete with granular camera motion controls and in-video object editing. Outputs can still be inconsistent, but its editing flexibility makes it a potent competitor in AI video.
Leonardo AI just added a new Motion feature that lets users turn static images directly into animated videos. Creators can adjust the motion strength to control the effect, with each generation costing 25 credits. Why it matters: With free accounts getting 150 daily credits, it gives creators an easy way to produce short video clips without specialized software.
Google just introduced VideoPoet, a multimodal model that ditches the prevailing diffusion approach in favor of a large language model architecture for video generation. It produces coherent 10-second clips featuring large motion and minimal artifacts, while handling text-to-video, image animation, and audio within a single unified model. It suggests LLMs could become a serious contender against diffusion in coherent video synthesis.
Midjourney just launched version 6 for public testing. It is not set as the default model yet, so users must manually select v6 in /settings or append --v6 to their prompts. Early users note that generation is running quite slowly right now.
