Runway just debuted Praxis-1, an open-weight world action model that adapts to new robots with light fine-tuning, alongside an end-to-end campaign agent called Runway Ads. Co-CEO Cristóbal Valenzuela warned that the US is still obsessed with LLMs while China pours token usage into video models, predicting live physical AI deployments are 12 to 18 months away.
Developers just open-sourced Mosaic Chunk, a memory system that helps AI video generators track objects and scenes long after they leave the frame. By selectively storing key features instead of every single frame, it maintains visual consistency across cuts. The release includes local-ready models based on MiniMax H3 and Ling World Infinity.
Tavus just unveiled Griffin, a full-duplex human interaction model that holds real-time video calls with lifelike expressions and gestures. Unlike turn-based bots, it watches and listens simultaneously, nodding mid-sentence or reacting to pauses, tricking 48% of live study participants into thinking it was human. Griffin-Lite is currently available to early testers.
Runway just launched Runway Ads, an autonomous engine that generates video and image creatives, publishes them directly to Meta, Google, and TikTok, and uses incoming performance data to generate the next batch. In internal testing, Runway scaled its weekly ad volume from 77 to 900 while cutting cost per subscriber by 41%.
Researchers just open-sourced Terra, an AI system that teaches simulated bodies to navigate complex terrain using realistic musculoskeletal mechanics rather than simple joint controls. Trained on over 9 hours of motion data, it walks across slopes and steps, and can be fine-tuned for jumps and rolls under the Apache 2.0 license.
Nvidia and the University of Waterloo just introduced PixelUMM, an open-source model that handles image and video understanding and generation directly in raw pixel space without a VAE or vision encoder. Built on Qwen3-8B, it runs all four multimodal tasks through a single decoder-only Transformer. Why it matters: Bypassing latent compression cuts down pipeline complexity and proves a unified raw-pixel approach is entirely feasible.
NVIDIA just introduced SoL-Refiner, an AI model that transforms low-resolution video drafts into crisp 4K outputs using a single denoising step. It works across different base generators, slashing MiniMax H3 generation time by 27x down to under six seconds. While it supercharges high-res rendering, it does alter or reimagine some fine details from the original clip.
InSpatio just open-sourced InSpatio-World 1.5, a spatial model that generates roaming videos from a single image, panorama, or video based on custom camera paths. It maintains spatial consistency and motion across time, and at under 6GB based on Wan 1.3B, it fits easily on consumer GPUs.
Researchers just open-sourced PDMD, a video distillation technique that shrinks MiniMax-H3 generation from 30 steps down to just four. By curbing misleading critic feedback during training, it keeps motion smoother and visual details sharp instead of muddying the frame. Fast open-source video generation is getting shockingly usable.
Google just launched Gemini 3.8 Live with Live Avatar, bringing real-time visual personas to its conversational AI in Gemini Enterprise. The system pairs near real-time video generation with speech, delivering fluid turn-taking and accurate lip-syncing across 97 languages. It can even call background tools without interrupting active dialogue.
Google just opened Google Vids to all personal and Workspace accounts at no cost, powered by its Gemini Omni 1.1 Flash model. Creators can generate 1080p HD scenes, extend clips with consistent lighting and characters, and set exact durations down to the second. Every clip carries an invisible SynthID watermark, and a multilingual voiceover tool is coming next.
Researchers just unveiled ACTx486, an experimental system that transforms pre-recorded videos into interactive media you can actually talk to. Viewers can interrupt the host with questions, request visual diagrams, or alter scene details before the clip seamlessly resumes. Full interactions still take minutes to generate, so the current demo is pre-rendered while the team aims for sub-six-second latency.
Google just released six creative workflow tools inside Google Flow, built in partnership with professional designers and filmmakers. The lineup includes multi-track sound effect generator Mondo Sónico, real-time 3D texture mapper Surface, and SwissFlow Studio for turning scripts into motion graphics. Creators can test them immediately or describe what they need to build their own tools.
YouTube just rolled out a suite of AI features for creators, bringing Gemini conversational editing straight into Shorts and the YouTube Create app. Creators can now chat with the assistant to trim clips and sync music to the beat, test up to three video cuts to see which hook works best, and lean on expanded voice detection to fight deepfakes. It basically turns YouTube Studio into an on-demand production partner.
Meta just rolled out a major update to its Meta Ray-Ban Display smart glasses. The glasses now offer landmark-based walking directions—telling you to turn at recognizable spots like a yellow house instead of street names—along with cycling and transit routes. A new Hologram feature creates a photorealistic digital avatar driven entirely by your voice for WhatsApp calls, eliminating awkward angles while cooking or multitasking. The update also brings Dolby Atmos recording and expands to five new countries.
PixVerse just launched PixVerse R2, a real-time world model built for interactive, generative video media. The system combines multimodal understanding with responsive audiovisual generation, letting you step into evolving scenes and reshape characters or plotlines on the fly while maintaining a coherent state.
Tavus just released Phoenix-4.5, a real-time human rendering model that cuts audio-to-video latency to 134ms. Instead of pasting a 512×512 face onto static footage, it renders the full upper body in a single pass with synchronized head, shoulder, and torso movements, plus zero-shot rendering in about a minute.
Tencent researchers just introduced WorldCrafter, an interactive video world model with an implicit 3D-aware memory that remembers scene structure when the camera turns away and revisits. It turns text prompts or single images into minute-scale explorable environments, with distillation enabling real-time streaming. Basically, AI video worlds no longer forget what was behind you.
Researchers just open-sourced GAE, a visual generation framework that bakes 3D structure directly into a compact 128-channel latent state. It decodes RGB video, depth maps, and camera trajectories simultaneously from a single representation. The approach cuts camera trajectory error in half, and the entire model weighs under 12GB for local consumer GPUs.
ElevenLabs just launched Studio 4.0, turning its editor into an all-in-one AI production suite. Creators can now generate video, images, sound effects, music, and voiceovers across 32+ languages directly on the timeline without switching tools. It also adds Studio Agent, an AI co-editor that drafts rough cuts and syncs audio frame by frame.
