DiffSynth-Studio just released Qwen-Image-i2L, an open-source suite that transforms a single image directly into a custom LoRA model. Built on SigLIP2, DINOv3, and Qwen-VL, it provides variants ranging from a 2.4B aesthetic extractor to a 7.9B coarse content model. Why it matters: Creators no longer need large datasets or tedious fine-tuning runs just to capture a unique style or subject.
Google Labs just upgraded its Mixboard creative canvas with Nano Banana Pro, allowing users to turn raw brainstorms directly into high-fidelity presentations. The update also brings doodle-based targeted editing, PDF and selfie uploads, and multi-board project support, turning freeform moodboards into presentation decks seamlessly.
Google just expanded Gemini 3 Pro and its Nano Banana Pro image model to Search's AI Mode across nearly 120 countries and territories in English. AI Pro and Ultra subscribers can now toggle on "Thinking with 3 Pro" to tap into complex reasoning, dynamic interactive UI generation, and custom simulations directly in query results.
Kling AI just launched Kling O1, a unified creative engine designed to process text, image, and video inputs together. The new model kicks off the team's Omni launch week and aims to streamline multimodal generation for creators. It is currently rolling out with limited-time access for paid subscribers.
OpenAI just teamed up with NORAD to launch three holiday tools inside ChatGPT for the annual Santa tracking tradition. Families can turn photos into elf helpers, generate printable coloring pages from toy ideas, and create custom read-aloud stories, all accessible via NORAD's official website.
OpenAI and Google just slashed free usage limits for Sora and Nano Banana Pro after surging holiday demand overwhelmed their compute. Sora free users are now capped at six video generations a day, while Google reduced image generations to two daily and restricted Gemini 3 Pro access. Basically, running frontier video and image models is so expensive that even the tech giants are feeling the squeeze.
Luma AI just unveiled Terminal Velocity Matching (TVM), a single-stage training paradigm that promises a 25x speedup over standard diffusion models when training from scratch. It scales to over 10B parameters and generates high-fidelity images in just four steps instead of the usual 100. Why it matters: it directly tackles the massive inference costs currently weighing down generative media.
Black Forest Labs just launched the FLUX.2 visual model family, supporting up to 10 reference images for character consistency, 4MP editing, and legible text rendering. The release includes the open-weight 32B FLUX.2 [dev] alongside pro-grade APIs. Why it matters: High-end creative pipelines just gained unprecedented visual control without being locked into closed ecosystems.
NotebookLM just rolled out a batch of creative updates, headlined by a new Slide Decks feature that turns your sources into visual stories alongside single-image infographics. The team also bumped custom persona limits from 500 to 5,000 characters and teased an upcoming Thinking UI.
Meta just introduced WorldGen, a research system that builds fully navigable, interactive 3D environments from a single text prompt in minutes. Unlike existing tools that break down when you walk a few meters away, it generates 50x50-meter textured spaces compatible with Unity and Unreal. It remains a research preview for now.
Midjourney just rolled out an early version of its new "Style Creator" tool. The feature lets creators explore and craft brand-new visual aesthetics without using any text prompts, offering a peek at the future of generative design.
Google just launched Nano Banana Pro, an upgraded image generation and editing model built on Gemini 3 Pro. It can pull real-time search data to build accurate infographics, render crisp multilingual text in full paragraphs, and blend up to 14 reference images while keeping faces consistent. Free users can test it with limited quotas inside the Gemini app.
MIT engineers just trained an AI to operate CAD software exactly like human designers, turning 2D sketches into 3D models. By learning from VideoCAD, a new dataset of over 41,000 screen-recorded UI actions, the agent mimics every click and drag needed to build objects. It sets the stage for an autonomous CAD co-pilot.
Meta just dropped SAM 3D, a new model family that reconstructs detailed 3D objects, scenes, and human bodies straight from flat 2D images. Split into Objects and Body models, Meta is open-sourcing the checkpoints and inference code while launching a web Playground. It bridges a notorious data gap in 3D modeling and is already powering virtual staging on Facebook Marketplace.
Meta just expanded its computer vision lineup with SAM 3 and SAM 3D. SAM 3 lets you track and segment objects in images and video using natural text prompts, while SAM 3D reconstructs 2D photos into workable 3D objects and human avatars. Both models are available to test in a new interactive playground, with SAM 3 code and weights fully open-sourced.
Midjourney just launched customizable user profiles, giving creators a dedicated space to showcase their best generations with custom banners, bios, and social links. To seed the community, anyone who completes their profile and spotlights at least 8 creations within 24 hours gets 5 free Fast GPU hours.
Replit just launched Design Mode, powered by the new Gemini 3 model to turn plain English prompts into polished, interactive mockups and static sites in under two minutes. You can preview clickable layouts right away and hit Convert to App whenever you're ready to add real backend code.
ElevenLabs just launched ElevenLabs Image & Video in beta, turning its voice platform into an all-in-one multimodal creative suite. Users can now generate visuals using models like Sora, Veo, Kling, and Flux, sync them with voiceovers and sound effects, and polish everything inside its Studio editor. It is a big push toward unifying visual and audio creation in a single workflow.
Google just partnered with luxury fashion brand Golden Goose to bring generative AI customization to over 40 stores worldwide. Shoppers can use Gemini image generation on a Pixel 10 to dream up personalized motifs, which in-store artisans then hand-paint directly onto sneakers. Why it matters: It turns consumer AI into a tactile, bespoke luxury experience rather than just another chatbot interface.
Fei-Fei Li, Yann LeCun, and DeepMind are all heavily betting on "world models," but they basically mean three completely different architectures. World Labs' Marble delivers interactive 3D Gaussian splats, LeCun's JEPA builds an internal predictive latent brain, and DeepMind's Genie 3 and SIMA 2 generate real-time video simulators. Why it matters: The industry is packaging front-end assets, robot gymnasiums, and core machine cognition under a single marketing buzzword.
