Index
Discover archive
Every Discover story, newest first.
No Discover articles yet.
No Discover articles with this tag.
-
Magnific's Move makes animated titles and charts from a typed description, then changes them when you ask
-
ElevenLabs says its new AI sound maker gets more of a busy scene's sounds from one written request
-
Pika's new AI video maker creates the sound with the picture, and its files are free to download
-
AssemblyAI now makes AI voices, built to keep callers from hanging up on phone bots
-
Liquid AI's small models are built to run on phones, and one can see, hear and talk back
-
Google DeepMind's Genie 4 makes worlds you can explore, with their own sounds and characters who talk
-
Runway's App Studio turns a typed description into a custom app for making video and images
-
Ideogram's new AI makes logos and art you can resize without blur and edit piece by piece
-
Free-to-download AI from VAST repaints blocky games and camera video in new styles, live
-
Skywork's free AI repaints a blocky video game in detail as you play
-
Meshy's new AI tool lets users circle one part of a 3D model and change it by typing
-
Unitree's app plans group dances for its humanoid robots, and its AI can draft them from typed requests
-
ElevenLabs' new AI writes music for a video and adds sound effects timed to the action
-
HeyGen says its AI can make a copied voice fit the scene, from a studio to a car
-
Fish Audio's AI voice is built to take direction like an actor, with files free to download
-
Google says its AI turned a cornhole video into moves a robot used to sink seven straight throws
-
In Nature Will's videos, robot hands fold wontons with AI trained on a million hours of hand video
-
A twin of Europe's upcoming Mars rover practices the drive the real one gets only one chance at: rolling off its lander
-
Vidu says its new AI video model can keep characters' looks and voices consistent across a scene
-
Magnific says its new AI image tool makes an art director's calls on framing, style and light for you
-
Kling says its next AI video tool can make 30-second scenes with sound and follow shots you plan
-
Google tested its medical chatbot on real patients, and the doctors watching never had to stop a chat
-
Google's new Playground lets people make a video game just by typing what they want
-
ElevenLabs' new AI voice can laugh, whisper and speak other languages in the same voice
-
Boston Dynamics gives its Atlas robot a four-finger hand built to use power tools
-
Runway's Praxis-1 aims to give robots a head start by learning from video
-
Sparse-WAM helps robot AI that predicts what its moves will do run about twice as fast, its authors say
-
Beihang and Peking University's RTP lets robot AI revise its video plan instead of starting over
-
InternRobotics researchers present InternW0-Delta, a robot AI that predicts what its moves will do
-
Nanjing University's AD-WM helps robots tell which move will work best
-
USC-led Rolling-WAM helps robots plan ahead and react faster
-
Stanford and Tel Aviv University researchers train AI from scratch with no real-world data
-
Shanghai AI Lab's InternW0 helps robots plan ahead at the lab bench
-
Tencent's GAE helps AI video keep 3D scenes in shape as the camera moves
-
Xiaomi's CodeMidas turns open-source code into 5,545 practice tasks for AI coding agents
-
Nankai University's DeformSmith builds soft virtual objects for robots
-
NUS, Tsinghua and partners release OpenWAM, a free robot AI built on video
-
X-Humanoid's Pelican-Sim 1.0 predicts what a robot's camera will see as it moves
-
XPENG's XPACE teaches its IRON humanoid new chores from videos of people
-
Odyssey says its Odyssey-3 AI can drive cars, steer robots and play games
-
Google's Dream-RSI improves how AI agents search for solutions by replaying their past attempts
-
Robbyant's LingBot-World turns AI video into a world you can play
-
Suno launches its v6 AI song models, developed with Warner Music Group, BMG and Believe
-
UIUC and Cornell's FIRE3D turns an everyday video of a room into editable 3D objects in under a minute
-
Alaya Lab's WorldSculpt rebuilds cluttered scenes as hundreds of separate, editable 3D objects
-
Mila researchers say trial-and-error training helps AI models most on easy problems, and propose a fix
-
Zhongguancun Academy releases ZGCM-1, a small, fully open AI model for math and for searching on its own
-
IBM releases a free Granite forecasting model that works on new data without extra training
-
Hugging Face and Earthmover publish a guide to running AI weather models anyone can download
-
Technical University of Munich paper argues AI that sees and reads should keep language at its edge, not its core
-
H Company releases NeoMME, small free AI models that handle text and page images together for document search
-
Anthropic says its Claude AI agents turned Fermat's Last Theorem into a computer-checked proof in 11 days
-
OpenAI says it has reached its goal of an automated research intern, with people still steering
-
Hugging Face survey of 13 AI labs finds each agent training attempt now gets its own computer
-
OpenAI says Astra is its first model to reach its Critical level for cybersecurity risk
-
Google releases Gemini 3.8 Flash, plus a Cyber version that only vetted defenders can use
-
xAI retires eight Grok models for developers, quietly rerouting most of them to grok-4.3
-
Tencent open-sources AuK, an AI model that creates and edits speech from plain-language instructions
-
Anthropic releases Claude Fable 5.1 to all customers and keeps its less-restricted twin, Mythos 5.1, for vetted groups
-
DeepSeek releases V4.1-Flash, a free-to-download AI model that reads images and text
-
M-A-P and HKUST release YuE2, a song generator that writes the score before the song