Jonathan Jørgensen 💙💛

@hxiao Cool idea :-)

small: single query: 7ms. Peak throughput: 25.3K tok/s. nano: single query: 2.9ms. Peak throughput: 98.7K tok/s. source github.com/jina-ai/jina-g…

now that jina-embeddings-v5-nano and v5-small are both super MLX-optimized, i can finally build jina-grep. back in 25/3 i said future embeddings would be more grep-like and less heavy-lifting, so i always wanted to do this - no time tho, until i saw @karpathy's tweet about cli, no reason to postpone it further

seedance 2.0 is the only model make me so scared literally every job in film industry is gone, you upload a script, it generates scenes (not just clips) with vfx, voice, sfx, music all nicely edited, we may not even need editors anymore and now I understand why it’s not available outside china there are features that feel so illegal, you upload screenshots or storyboard frames from any movie, and it generates full scenes that feel like from the original movie, but its different and sometimes looks better another feature is crazier, you upload any film clips, and you can just… edit anything.. like swap characters, add vfx, change bg, color grading.. at this point maybe the only thing we still need is screenwriters because the stories AI write are still cliché but.. the machine is learning faster and faster this feels like the quiet end of traditional film industry and the beginning of something we dont know

Seedance 2.0 is dropping soon I'm calling it, it will break the internet, 100%. It will be SOTA, with Kling 3.0 just released, this will leap yet another step forward. This is the official threshold, the point of no return. More examples below 👇

Video message from Federal Reserve Chair Jerome H. Powell: federalreserve.gov/newsevents/spe…

@ThierryBreton Group think

Is McCarthy’s witch hunt back? 🧹 As a reminder: 90% of the European Parliament — our democratically elected body — and all 27 Member States unanimously voted the DSA 🇪🇺 To our American friends: “Censorship isn’t where you think it is.”

@stevemushero @sobedominik I’ve been using them since 2014, when they seemed to be the only company with a reliable USB hub

@sobedominik Were they really ever a cheap seller? I thought they started off fairly high-end from the beginning (though hard in early batteries).

Anker might be the best brand turnaround story nobody talks about. 2011: Cheap Chinese battery seller on Amazon. Looked like every other Alibaba dropshipper and everyone made fun of you for owning one. 2025: Competing with Apple/Belkin, exceptional design, pioneering GaN charging tech. All my power banks, bricks and cables are from Anker. They made 60% of their team R&D instead of just outsourcing everything and actually listened to customer feedback. Basically proved you can build a trusted & high quality hardware brand from a "generic Chinese knockoff". Wild that more companies haven't really copied this playbook.

@toptopseb I tvivl om min er bedre når den siger 19 😬

Spotify wrapped mener at jeg følge min musiksmag er 70 år 😃

Contact Sheet prompting in Nano Banana Pro is getting a lot of buzz. I tried adapting it for a 'fashion style' shoot... I'm sold. Full workflow, prompts, and some tips below. Shout out @techhalla @firatbilal for the inspiration.

@LignoL23 @criccomini What is your experience for prod use? It says “experimental” on the site

@criccomini This is also possible with rclone rclone s3 serve rclone.org/commands/rclon…

Someone just pointed me at this: github.com/versity/versit… Seems neat.

Here's a tutorial on how to take footage and turn it into a fictive character using Kling O1 on @freepik

Kling Omni Launch Week Day 1: Introducing Kling O1 — Brand-New Creative Engine for Endless Possibilities! Input anything. Understand everything. Generate any vision. With true multimodal understanding, Kling O1 unifies your input across texts, images, and videos — making creation faster, smarter, and more effortless. Limited-Time offer available for subscribers. And stay tuned for more surprises! For the next 12 hours ONLY Follow, Like & Retweet to get 200 Credits — for everyone who participates! PLUS, 200 lucky winners will snag a *1-Month Standard Plan* All sent straight to your DM

Introducing our new frontier video model, Runway Gen-4.5. Previously known as Whisper Thunder (aka) David. Gen-4.5 is state-of-the-art and sets a new standard for video generation motion quality, prompt adherence and visual fidelity. Learn more below.

Introducing Claude Opus 4.5: the best model in the world for coding, agents, and computer use. Opus 4.5 is a step forward in what AI systems can do, and a preview of larger changes to how work gets done.

Today we’re excited to unveil a new generation of Segment Anything Models: 1️⃣ SAM 3 enables detecting, segmenting and tracking of objects across images and videos, now with short text phrases and exemplar prompts. 🔗 Learn more about SAM 3: go.meta.me/591040 2️⃣ SAM 3D brings the model collection into the 3rd dimension to enable precise reconstruction of 3D objects and people from a single 2D image. 🔗 Learn more about SAM 3D: go.meta.me/305985 These models offer innovative capabilities and unique tools for developers and researchers to create, experiment and uplevel media workflows.

Amazing work!

In two years, the level of detail you can capture in a 3D Gaussian Splat has skyrocketed. In this scene on SuperSplat, I can zoom right in on a certificate and the text is perfectly legible. Imagine where we will be in another 2 years. 🪄

We are just scratching the surface of precise control over AI video generation. MotionStream unlocks real-time video with interactive motion controls. You can interactively generate video based on motion inputs (like drawn trajectories, camera movements, or motion transfer). 29fps generation w/ 0.4 second latency on a single H100, oh and infinite length streaming. So much more intuitive than text prompts. Almost feels like performance capture - painting motion paths and watching AI generated video follow those paths instantly. Amazing work by Xun Huang et al. Feels like we're on the cusp of a new type of creation engine. And let me tell you - it ain't all nodes :-)

Cool new work on real-time video generation with motion control. MotionStream enables sub-second latency streaming on a single GPU. You can also control what parts of an image do & don't move, like the video below. 👏 to @xxunhuang, @jhshin2000, and team!

@narendramodi What the heck

Thank you, my friend, President Putin, for your phone call and warm wishes on my 75th birthday. We are committed to further strengthening our Special and Privileged Strategic Partnership. India is ready to make all possible contributions towards a peaceful resolution of the Ukraine conflict. @KremlinRussia_E

Various animals, mid-flip. Seedream 4 handling these just fine. > IMG_3984.CR2 a [animal] mid flip