Runway Gen-4 is here: Taking AI video creation to the next level

Hi AI lovers! Here’s what you need to know about AI today:

👉 Runway’s next-generation AI video generator, Gen-4, is here

👉 Amazon released Nova Act for agentic web automation tasks

👉 OpenAI’s new image generation tool is now available to free users

and many more!

📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!

Read time: 5 minutes

What’s going on: Runway, a New York-based AI startup, released Gen-4, a video generation model that pushes the boundaries of what AI can achieve in dynamic, high-quality video output. Backed by giants like Google, Nvidia, and Salesforce, the company says Gen-4 outclasses its earlier Gen-3 Alpha model by delivering sharper 720p clips, lasting 5 to 10 seconds, with better motion and consistency in characters, settings, and objects. Using text prompts and reference images, it’s built to maintain narrative flow and stylistic unity, making it a practical tool for creators working on everything from short stories to product demos.

What does it mean: The model’s strength lies in its ability to follow detailed instructions and incorporate visual cues, ensuring the output stays on track across frames. Runway claims Gen-4 surpasses competitors like OpenAI’s Sora and others in prompt accuracy and scene coherence.

More details: 

• Runway Gen-4 is available now to paid users, both individuals and businesses.

• With this launch, Runway is aiming for a lead in the AI video race against OpenAI, Google, and others, eyeing a $4 billion valuation and $300 million in yearly revenue.

• Many people in the creative and film-making industry are panicking. A 2024 animation guild report flags tools like Gen-4 as a threat, projecting over 100,000 US creative jobs could vanish by 2026.

• Want to watch many cool example videos generate by Gen-4 and try it yourself? Visit this page.

What’s going on: Amazon has introduced Nova Act, a new AI agent capable of independently controlling a web browser to perform basic tasks. Nova Act is designed to handle actions like ordering food or booking dinner reservations, showing off its potential as a practical digital assistant. Developed by Amazon’s San Francisco-based AGI lab which is led by former OpenAI researchers David Luan and Pieter Abbeel, this agent is part of a broader effort to compete with AI offerings from OpenAI and Anthropic.

What does it mean: Nova Act is supposed to enhance Amazon’s upcoming Alexa+ upgrade, integrating generative AI to make the voice assistant more autonomous and capable. The current version, labeled a research preview, is less refined but still demonstrates strong performance, scoring 94% on Amazon’s ScreenSpot Web Text test, outpacing OpenAI’s CUA (88%) and Anthropic’s Claude 3.7 Sonnet (90%).

More details: 

• Alongside the agent, Amazon released the Nova Act SDK, a toolkit that enables developers to create their own agent-based applications, expanding the technology’s reach beyond Amazon’s ecosystem.

• While it hasn’t been tested against broader benchmarks like WebVoyager, Amazon claims superiority in reliability for simple tasks.

• The SDK allows developers to define workflows where human intervention can occur, addressing a key challenge in early AI agents that is consistency across varied domains.

• The launch places Amazon in a crowded field, racing against OpenAI’s Operator and Anthropic’s Computer Use, Manus AI, ByteDance’s UI-TARS as well as similar efforts from Google and others.

• Early AI agents from competitors have struggled with speed, independence, and error rates, issues Amazon aims to sidestep with Nova Act’s design.

• Want to experiment with it? Nova Act is available via nova.amazon.com.

💰 OpenAI announced a $40 billion funding round led by SoftBank, valuing the company at $300 billion post-money, to further AI research, scale compute infrastructure, and enhance tools for ChatGPT's 500 million weekly users.

🎙 Amazon launched Alexa+ recently, but many of the promised "agentic" AI features like ordering takeout, generating stories, visually identifying people for chore reminders, and brainstorming gift ideas are missing or only available to a limited number of users, due to ongoing issues with the underlying AI infrastructure.

📷 OpenAI has made its new GPT-4o-powered image generator available to all users.

🧾 ChatGPT's new image generator, excels at creating realistic text within images, leading to concerns about the ease of generating convincing fake receipts for fraudulent purposes.

🔔 Apple is rolling out Priority Notifications, a new feature of Apple Intelligence that highlights time-sensitive notifications, along with other updates to Image Playground and the Mac, as Apple Intelligence expands to EU iPhone and iPad users and to the Apple Vision Pro headsets.

❌ OpenAI has temporarily disabled Sora video generation for new accounts, due to ongoing capacity issues following the launch of its viral image generation feature.

AI + Travel planning

Airbnb - Senior Staff Machine Learning Scientist

Wiliot - Data Scientist

Captions - Software Engineer, Video Processing

OpenAI - Software Engineer, Developer Productivity

Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!