In partnership with |
 |
|
|
Your Competitors Just Hired an Entire GTM Team. No Humans. |
|
AI agents are doing the work of full GTM teams. Pipeline, content, customer service. The startups using them aren't waiting for headcount approvals. |
Get the free Practical Guide to Agentic GTM for Startups and see exactly where to start. |
Unlock the Playbook |
AI video just crossed a line nobody expected this soon: generation is now faster than playback. fal's new H3 Max model turns a prompt into a 5-second clip with synced audio in under 3 seconds, and creators on X are already calling it a real inflection point for the medium. |
The timing lines up with a wave of other moves across the industry: Nvidia is reportedly closing in on its biggest acquisition ever, and Google just handed its video model a serious upgrade in editing control. If render times keep shrinking toward zero, how long before AI video feels less like generation and more like live directing? |
Today in AI: |
fal's H3 Max generates video faster than real time Nvidia reportedly nears $12.9B Hugging Face acquisition Google's Gemini Omni Flash adds precise editing controls |
|
|
|
What's new? fal released H3 Max, a post-trained version of MiniMax's open-weight H3 model that generates a 5-second video clip with synced audio in under 3 seconds - the first time generation speed has beaten playback time on a frontier-quality video model. |
What matters? |
H3 Max runs roughly 35 times faster than the official MiniMax H3 endpoint, according to fal's own testing, while still ranking #1 for image-to-video quality on independent leaderboards. Every clip ships with native synced audio - dialogue, music, and sound effects generated in the same pass as the video, not bolted on afterward. Anyone can test it on fal's playground with five free 15-second generations a day, no coding required. |
Why it matters? |
When generation is faster than playback, AI video stops being a render-and-wait chore and starts working like a live creative loop: prompt, watch, tweak, regenerate. That shift turns AI video into something closer to an interactive tool than a batch process, and it hands every other lab a new speed target to chase. |
|
GUIDE |
|
|
|
What's new? Nvidia has reportedly agreed to buy Hugging Face, the open-source hub that hosts most of the world's shared AI models, for $12.9 billion - a deal that would be Nvidia's largest ever if it closes. |
What matters? |
The reported price is more than 1.8 times the $6.9 billion Nvidia paid for Mellanox in 2020, its previous largest acquisition. Nvidia was already an investor in Hugging Face, having backed the startup in an earlier funding round before this deal emerged. Neither company has confirmed the deal, and TechCrunch reports no contract has been signed yet, so talks could still fall apart. |
Why it matters? |
Hugging Face is where developers pick the open models that end up running on Nvidia's chips, including the image and video generators that power much of today's creative AI. Owning that layer would let Nvidia shape which models gain traction - and how much of its own compute they end up needing. |
|
SPONSORED BY SUPERHUMAN AI |
Go from AI overwhelmed to AI savvy professional |
|
AI will eliminate 300 million jobs in the next 5 years. |
Yours doesn't have to be one of them. |
Here's how to future-proof your career: |
Join the Superhuman AI newsletter - read by 1M+ professionals Learn AI skills in 3 mins a day Become the AI expert on your team |
Start learning AI now |
|
|
|
What's new? Google rolled out Gemini Omni 1.1 Flash, a video generation and editing model that lets creators extend scenes, set exact start and end frames, and upscale outputs to 4K - all inside Google AI Studio, Flow, and the Gemini app. |
What matters? |
Scene extension analyzes up to 10 seconds of existing footage and stretches a clip in 10-second increments, up to 40 seconds total, while keeping character and lighting consistent. A new 360p draft mode generates lightweight previews 60% faster and at a third of the cost of full-resolution renders, so creators can iterate before committing to a final render. Runway is already building the model into its own platform, with Chief Creative Officer Jamie Umpherson saying it "fits naturally into how people already use Runway." |
Why it matters? |
Precise frame control and scene extension move Google's video model from a demo generator toward an actual editing tool, the kind of control creators need to fit AI video into real production pipelines. Partnering with Runway on day one signals Google wants this model embedded in professional workflows, not just standalone apps. |
|
Everything else in AI |
Midjourney explains how its Personalization system builds a custom taste profile from ranked image pairs, then applies it to every prompt with the --p parameter. |
Hugging Face unveiled Microduck, a $399 open-source robot duck that ships with seven pre-trained moves and can be taught new tricks through reinforcement learning. |
The EU began enforcing AI Act rules that require clear labels on AI-generated images and video shown to EU audiences, with fines up to €15 million for violations. |
Complete AI Bundle packs prompt libraries, courses, and templates into one download for readers who want to turn daily AI use into paid work. |
|
Essential AI Guides - Reading List: |
|
|
Let us know! |
What did you think of today's email?
Before you go, please give your feedback to help us improve the content for you!
|
|
|
Work with us |
Reach 100k+ engaged Tech Professionals, Engineers, Managers and decision makers. Join brands like MorningBrew, HubSpot, Prezi, Nike, Ahref, Roku, 1440, Superhuman, and others in showcasing your product to our audience. Get in touch now → |
