Скачать с ютуб The Weekly AI Roundup: Runway Gen 4, OpenAI Academy, Midjourney 7, Wave 6, Higgsfield AI Video Model в хорошем качестве

The Weekly AI Roundup: Runway Gen 4, OpenAI Academy, Midjourney 7, Wave 6, Higgsfield AI Video Model

Runway launched Gen 4 and firmly stamped its authority back on the video generation space. Of course, the fourth generation model looks incredible, but it's the control of the output that is likely to be the biggest selling point. The launch video explains how both character and scene consistency can be achieved across multiple outputs, allowing Runway to really look like a serious filmmaking tool. Google CEO Sundar Pichai surprised everyone at the weekend by announcing that their state of the art Gemini 2.5 Pro model would be made available to everyone for free. It's quite incredible that all users can now access this model at no cost. Google is winning a lot of fans with their recent work. Isomorphic Labs, a company spun off from Google's DeepMind, has secured $600 million of funding to advance AI driven drug discovery. The funding will be used to accelerate the development of AI designed pharmaceuticals. Isomorphic Labs plans to start clinical trials by the end of 2025, focusing on using AI to address all diseases and potentially transform the pharmaceutical industry. OpenAI managed to successfully raise another $40 billion in funding in a new round led by SoftBank. This set a new record for the largest private tech financing ever and values OpenAI at a staggering $300 billion. OpenAI also announced plans to launch its first open source model since GPT2. Sam Altman posted on X "We are excited to release a powerful new open weight language model with reasoning in the coming months and we want to talk to devs about how to make it maximally useful." Krea announced that they are introducing natural language image editing into their workflows via the Gemini models and they showed off a whole new UI which includes a brand new 3D tool. Unitree Robotics has created a new robotic hand with 20 degrees of freedom and it can perform some really incredible moves. The Dex 5 dextrous hand has 16 active and 4 passive joints, giving this hand incredibly smooth, precise and agile movements. OpenAI has launched the OpenAI Academy, a free educational platform designed to democratise AI education. This is likely to be an invaluable resource for anyone looking to learn AI as they will get high quality resources from one of the industry leaders and it's free. Anthropic has launched 'Claude for Education' a version of their AI model that is tailored for higher education. The new 'Learning Mode' is designed to promote critical thinking among students by integrating AI into in a manner that encourages independent thought rather than just providing direct answers. Windsurf has released Wave 6. This update allows users to deploy apps directly to the Internet with a single click using the built in Cascade tool. There are many other updated features including commit, message generation and improved MCP support. Lindy launched a new feature called 'Agent Swarms' which allows for the simultaneous operation of hundreds of AI agents to perform tasks in parallel. The Tesla Optimus robot took a stroll around the Gigafactory today and impressed everyone with its new swagger. Luma Labs added camera controls to their video generation, allowing the user to select one of 20 different camera motion presets. Krea have added a great new restyle option that allows the user to upload a source video and then have it reimagined in a completely different style. Pika Labs has introduced Pico Frames. This allows the user to add up to five keyframes to produce 25 second videos. This can be used for more precisely animating a scene or for producing some amazing transformations between different images. Higgsfield AI is a video generation model that puts the camera motion at the heart of the process. In fact, it's the very first step you take in the process. Choose from one of many different camera motions, add an image, add a prompt, and produce some of the coolest looking AI videos around. The once undisputed king of AI image generation, Midjourney, launched its latest model V7 into an alpha test phase. Midjourney is famous for its ability to create ultra realistic images, and this is still the area where the new model seems to excel. We've seen some stunning early examples, however, it seems that the creative process might actually be the thing that does help Midjourney stand out from the crowd. The new creative experience allows the user to make rapid iterative edits to the images created, and with voice inputs, this can be a literal creative conversation. OpenAI has announced some changes to its model release schedule. They're now going to release an o3 model that previously they had suggested would be skipped, and also an o4-mini version. Sam Altman explained on X that, "we are going to be able to make GPT5 much better than we originally thought. We also found it harder than we thought it was going to be to smoothly integrate everything".