Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

damn. spatial intelligence can be FAST asf! "D4RT can continuously understand what's moving while running 18x–300x faster than previous methods - processing a 1-minute video in roughly 5 seconds on a single TPU chip."

Bilawal Sidhu

105,018 subscribers

33,587 views • 7 months ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Grok-1 support in gpt-fast at faster(?) than anyone else has reported so far. 75 tok/s for a 300B+ parameter model on an 8xA100 node. If I understand correctly, ColossalAI reported 15 seconds to generate 100 tokens. gpt-fast takes 4.2 seconds to generate *400* tokens.

Grok-1 support in gpt-fast at faster(?) than anyone else has reported so far. 75 tok/s for a 300B+ parameter model on an 8xA100 node. If I understand correctly, ColossalAI reported 15 seconds to generate 100 tokens. gpt-fast takes 4.2 seconds to generate 400 tokens.

Horace He

54,137 views • 2 years ago

A few seconds from prompt to video — that's faster than Twitch streams live. H3 Max on MiniMax Design, post-trained by fal. The speed-and-cost combo is what really stands out. You can iterate in seconds, not minutes.

A few seconds from prompt to video — that's faster than Twitch streams live. H3 Max on MiniMax Design, post-trained by fal. The speed-and-cost combo is what really stands out. You can iterate in seconds, not minutes.

KICHOCHEZI

311,115 views • 13 days ago

(1/n) 🚀 With FastVideo, you can now generate a 5-second video in 5 seconds on a single H200 GPU! Introducing FastWan series, a family of fast video generation models trained via a new recipe we term as “sparse distillation”, to speed up video denoising time by 70X! 🖥️ Live demo: (Thanks to @gmicloud for the support!) 🔗 Blog: 🔓 We fully open-source our models, code, and data with Apache-2.0 licenses

(1/n) 🚀 With FastVideo, you can now generate a 5-second video in 5 seconds on a single H200 GPU! Introducing FastWan series, a family of fast video generation models trained via a new recipe we term as “sparse distillation”, to speed up video denoising time by 70X! 🖥️ Live demo: (Thanks to @gmicloud for the support!) 🔗 Blog: 🔓 We fully open-source our models, code, and data with Apache-2.0 licenses

Hao AI Lab

78,660 views • 1 year ago

Video quality will probably be garbled but I updated the lightbox on my site so that the maps can be viewed in fullscreen & zooms in on a single click rather than double. Should be a lot more intuitive & fit nicer on a 2nd monitor.

Video quality will probably be garbled but I updated the lightbox on my site so that the maps can be viewed in fullscreen & zooms in on a single click rather than double. Should be a lot more intuitive & fit nicer on a 2nd monitor.

REƎMR

95,116 views • 9 months ago

NVIDIA open-sourced a 600M model that transcribes 40 languages in real-time at 80ms latency and it costs $0. that's faster than you can blink. across mandarin, arabic, hindi, portuguese, tagalog, whatever,from a SINGLE checkpoint. → 17x more concurrent streams than buffered ASR on the same H100. → punctuation + capitalization built-in. no post-processing. → runs on your own GPU. no API bill 100% Open Source.

NVIDIA open-sourced a 600M model that transcribes 40 languages in real-time at 80ms latency and it costs $0. that's faster than you can blink. across mandarin, arabic, hindi, portuguese, tagalog, whatever,from a SINGLE checkpoint. → 17x more concurrent streams than buffered ASR on the same H100. → punctuation + capitalization built-in. no post-processing. → runs on your own GPU. no API bill 100% Open Source.

Superman

104,663 views • 1 month ago

BIOS Dev Update: Fast Chat, Thinking Traces & More What's New: • Fast Chat Mode: A faster pipeline for simple questions that switches to literature search only when needed. It runs in an agentic loop, searching as many times as it needs until it has enough context to give you a complete answer. • Thinking Traces on Screen: BIOS now shows its reasoning on the main loading screen while Deep Research is running, so you can follow along in real time.

BIOS Dev Update: Fast Chat, Thinking Traces & More What's New: • Fast Chat Mode: A faster pipeline for simple questions that switches to literature search only when needed. It runs in an agentic loop, searching as many times as it needs until it has enough context to give you a complete answer. • Thinking Traces on Screen: BIOS now shows its reasoning on the main loading screen while Deep Research is running, so you can follow along in real time.

BIOS

10,285 views • 5 months ago

AI video is moving beyond "generate a clip". We're entering the era of directing entire scenes. With MiniMax H3 on Magnific, you can combine: → Text → Images → Videos → Audio in a single workflow. Then control: → Character → Camera → Motion → Sound with far more consistency than a prompt alone. That's what makes this model interesting. 50% OFF until September 1.

AI video is moving beyond "generate a clip". We're entering the era of directing entire scenes. With MiniMax H3 on Magnific, you can combine: → Text → Images → Videos → Audio in a single workflow. Then control: → Character → Camera → Motion → Sound with far more consistency than a prompt alone. That's what makes this model interesting. 50% OFF until September 1.

Bishal Nandi

53,240 views • 29 days ago

AI video is moving FAST. Wan 3.0 on Magnific can now generate 30-second videos with sound, consistent characters, action, camera movement, voices, music, and effects in one call. This feels less like AI video generation and more like directing a full scene with a single prompt. Get up to 40% OFF until Sept 30:

AI video is moving FAST. Wan 3.0 on Magnific can now generate 30-second videos with sound, consistent characters, action, camera movement, voices, music, and effects in one call. This feels less like AI video generation and more like directing a full scene with a single prompt. Get up to 40% OFF until Sept 30:

Atul Kumar

69,450 views • 18 days ago

Tesla and SpaceX are about to need chips faster than the world can produce. The answer? Terafab. A chip factory in Texas so huge it will dwarf landmarks like the Pentagon and Apple Park, with a goal of producing more than 1 terawatt of compute every year. The point is to stop future Tesla and SpaceX growth from being limited by whoever can supply them. Elon says Terafab will eventually be worth more than any other building on Earth. Tesla SpaceX / Writer: Annette, Grok Imagine Designer: Janné

Tesla and SpaceX are about to need chips faster than the world can produce. The answer? Terafab. A chip factory in Texas so huge it will dwarf landmarks like the Pentagon and Apple Park, with a goal of producing more than 1 terawatt of compute every year. The point is to stop future Tesla and SpaceX growth from being limited by whoever can supply them. Elon says Terafab will eventually be worth more than any other building on Earth. Tesla SpaceX / Writer: Annette, Grok Imagine Designer: Janné

Mario Nawfal

135,609 views • 1 month ago

You can win $100,000 by bringing a scene from Homer’s The Odyssey to life with Grok Imagine! SpaceXAI has launched a filmmaking contest challenging creators to produce a compelling scene based on Homer’s ancient Greek epic while showcasing Grok Imagine's video and voice capabilities. Requirements: • A 3–5 minute video in English • At least 1 minute of dialogue • Grok Imagine must create the video, images and character voices • Other tools can be used for editing, music and sound effects Prizes: 🏆 1st: $100,000 🥈 2nd: $50,000 🥉 3rd: $25,000 Open to eligible U.S. residents aged 18 or older. The deadline is August 31, 2026.

You can win $100,000 by bringing a scene from Homer’s The Odyssey to life with Grok Imagine! SpaceXAI has launched a filmmaking contest challenging creators to produce a compelling scene based on Homer’s ancient Greek epic while showcasing Grok Imagine's video and voice capabilities. Requirements: • A 3–5 minute video in English • At least 1 minute of dialogue • Grok Imagine must create the video, images and character voices • Other tools can be used for editing, music and sound effects Prizes: 🏆 1st: $100,000 🥈 2nd: $50,000 🥉 3rd: $25,000 Open to eligible U.S. residents aged 18 or older. The deadline is August 31, 2026.

DogeDesigner

66,507 views • 26 days ago

I didn’t expect AI to make ad creation this fast. Just tested Pollo AI Marketing Studio to see what it can do. Tried making a sports streaming app ad from a single product image. No editing. No timeline. No production setup. Just picked a video format and it generated a full video ad in seconds. What stood out wasn’t just the output,, it was how simple the whole process is now. What used to take multiple steps and tools can now be done in a single workflow. Curious how others see this, would you use something like this for ads?

I didn’t expect AI to make ad creation this fast. Just tested Pollo AI Marketing Studio to see what it can do. Tried making a sports streaming app ad from a single product image. No editing. No timeline. No production setup. Just picked a video format and it generated a full video ad in seconds. What stood out wasn’t just the output,, it was how simple the whole process is now. What used to take multiple steps and tools can now be done in a single workflow. Curious how others see this, would you use something like this for ads?

Ai Girllie

18,849 views • 3 months ago

💡HunyuanVideo1.5 Update: We are now releasing the 480p I2V step-distilled model, which generates videos in 8 or 12 steps (recommended)! On RTX 4090, end-to-end generation time is reduced by 75%, and a single RTX 4090 can generate videos within 75 seconds. The step-distilled model maintains comparable quality to the original model while achieving significant speedup. For even faster generation, you can also try 4 steps (faster speed with slightly reduced quality). 🔗Check out the GitHub Repo:

💡HunyuanVideo1.5 Update: We are now releasing the 480p I2V step-distilled model, which generates videos in 8 or 12 steps (recommended)! On RTX 4090, end-to-end generation time is reduced by 75%, and a single RTX 4090 can generate videos within 75 seconds. The step-distilled model maintains comparable quality to the original model while achieving significant speedup. For even faster generation, you can also try 4 steps (faster speed with slightly reduced quality). 🔗Check out the GitHub Repo:

Tencent Hy

39,011 views • 9 months ago

A single product image can now become a cinematic ad in minutes. Created with "𝐏𝐡𝐨𝐭𝐨 𝐭𝐨 𝐕𝐢𝐝𝐞𝐨 𝐀𝐝𝐬" on the Pollo AI mobile app. What used to take a full creative pipeline can now be turned into a polished, conversion-focused video ad in one workflow, making it especially useful for ecommerce sellers who need to launch and test creatives fast. AI advertising is accelerating.... See comment for a Prompt 👇

A single product image can now become a cinematic ad in minutes. Created with "𝐏𝐡𝐨𝐭𝐨 𝐭𝐨 𝐕𝐢𝐝𝐞𝐨 𝐀𝐝𝐬" on the Pollo AI mobile app. What used to take a full creative pipeline can now be turned into a polished, conversion-focused video ad in one workflow, making it especially useful for ecommerce sellers who need to launch and test creatives fast. AI advertising is accelerating.... See comment for a Prompt 👇

Aneeza Ai

32,912 views • 3 months ago

Pothole detection on the road in real time using Ultralytics YOLO26! 🕳️ Manual road inspections are slow, costly, and hard to scale. With object detection, potholes can be identified directly from street-level images or video feeds, enabling faster and more consistent road condition monitoring. How I built this demo: ✅ Trained a segmentation model on a custom dataset. ✅ Generated mask contours for each pothole. ✅ Leveraged the onnx-exported model for faster processing. #Pothole #RoadDamage #AI

Pothole detection on the road in real time using Ultralytics YOLO26! 🕳️ Manual road inspections are slow, costly, and hard to scale. With object detection, potholes can be identified directly from street-level images or video feeds, enabling faster and more consistent road condition monitoring. How I built this demo: ✅ Trained a segmentation model on a custom dataset. ✅ Generated mask contours for each pothole. ✅ Leveraged the onnx-exported model for faster processing. #Pothole #RoadDamage #AI

Muhammad Rizwan Munawar

30,677 views • 5 months ago

$🚀 A better, faster co-folding-based binding affinity model. Predicting how tightly a drug candidate binds to its target is critical in drug discovery. It also requires massive computational resources. State-of-the-art models can take 20 seconds to a minute per prediction, impractical for the demands of large scale early-stage programs . 💠 Today, Recursion’s Valence Labs is releasing Nesso-1: the fastest open-source co-folding-based binding affinity model available. At 1 second per prediction, it’s roughly 20x faster than our previous collaboration on Boltz-2 while matching or surpassing its accuracy across public and internal benchmarks. By leveraging NVIDIA Healthcare cuEquivariance, we’ve been able to further accelerate both training and inference by an additional 2-3x. We look forward to continuing to improve Nesso-1 in collaboration with NVIDIA. Weights and code are fully open-sourced. The core architectural ideas behind Nesso-1 build on the insight that coarse-grained co-folding representations can match full-atom models for affinity prediction at a fraction of the cost. Nesso-1 is the first open implementation of this approach with no proprietary dependencies, trained entirely on public data, built to be reproducible and extensible. We’re already using Nesso-1 internally in active drug discovery programs. Fast, reliable affinity prediction at scale is foundational to the kind of autonomous design loops that define our vision for Autonomous Precision Design and Nesso-1 is a meaningful step toward that. 👉 Report: 👉 Github: 👉 HF:$

🚀 A better, faster co-folding-based binding affinity model. Predicting how tightly a drug candidate binds to its target is critical in drug discovery. It also requires massive computational resources. State-of-the-art models can take 20 seconds to a minute per prediction, impractical for the demands of large scale early-stage programs . 💠 Today, Recursion’s Valence Labs is releasing Nesso-1: the fastest open-source co-folding-based binding affinity model available. At 1 second per prediction, it’s roughly 20x faster than our previous collaboration on Boltz-2 while matching or surpassing its accuracy across public and internal benchmarks. By leveraging NVIDIA Healthcare cuEquivariance, we’ve been able to further accelerate both training and inference by an additional 2-3x. We look forward to continuing to improve Nesso-1 in collaboration with NVIDIA. Weights and code are fully open-sourced. The core architectural ideas behind Nesso-1 build on the insight that coarse-grained co-folding representations can match full-atom models for affinity prediction at a fraction of the cost. Nesso-1 is the first open implementation of this approach with no proprietary dependencies, trained entirely on public data, built to be reproducible and extensible. We’re already using Nesso-1 internally in active drug discovery programs. Fast, reliable affinity prediction at scale is foundational to the kind of autonomous design loops that define our vision for Autonomous Precision Design and Nesso-1 is a meaningful step toward that. 👉 Report: 👉 Github: 👉 HF:

Recursion

156,802 views • 1 month ago

Running out of memory used to be a common headache when running molecular simulations with neural network potentials. Not any more. Here Orb is simulating over half a million atoms on a single GPU (H200). This is a fully solvated COVID spike protein. (The trick is to specify brute_force_knn=False in the calculator.) This is more than an order of magnitude larger than some other universal models. I can get over 7,000 steps of molecular dynamics per day with it.

Running out of memory used to be a common headache when running molecular simulations with neural network potentials. Not any more. Here Orb is simulating over half a million atoms on a single GPU (H200). This is a fully solvated COVID spike protein. (The trick is to specify brute_force_knn=False in the calculator.) This is more than an order of magnitude larger than some other universal models. I can get over 7,000 steps of molecular dynamics per day with it.

Tim Duignan

106,630 views • 1 year ago

Churn above single digits will quietly bleed your business dry. If you're running an MRR offer, it should be 5% or less. Problem is, most people have no idea what's actually driving theirs up. So I put together a free Client Retention Playbook. (The 5-step playbook holding churn under 5% on my own recurring offer) Like this + comment 'CHURN' and I'll send it over. *Must follow so I can send the DM*

Churn above single digits will quietly bleed your business dry. If you're running an MRR offer, it should be 5% or less. Problem is, most people have no idea what's actually driving theirs up. So I put together a free Client Retention Playbook. (The 5-step playbook holding churn under 5% on my own recurring offer) Like this + comment 'CHURN' and I'll send it over. Must follow so I can send the DM

Jeremy Haynes

13,438 views • 8 days ago