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.

NVIDIA drop a 3B vision-language model for fast, high-quality visual grounding in real time accurately. - Parallel box decoding - 10× faster than Qwen3-VL - Trained on 138M queries/785M boxes - GUI, OCR, and document layout, dense detection - Open source Useful for computer-use agents and Physical AI,

Md Ismail Šojal 🕷️

57,104 subscribers

131,447 views • 4 days ago •via X (Twitter)

News & Politics Education Science & Technology

Anya Rossi• Live Now

Private livecam show

14 Comments

Md Ismail Šojal 🕷️4 days ago

Worth a look if you work on agents, robotics, or document AI. -

falky4 days ago

shit is 4 months old, fucking clickbait

Fajar M Reza4 days ago

Does the 10× speedup hold on GUI workloads, or mainly dense grounding?

Samurai.AI4 days ago

@skalskip92

SHK4 days ago

@grok find this model to download

Dylan Colquhoun4 days ago

Whats the optimal packing of 17 boxes in a square?

Derek Chia4 days ago

Parallel box decoding sounds like the real trick. How does it hold up on dense GUI screens versus Qwen3-VL?

CEO del Socialismo de Mercado🌹🕊️ 市场社会主义CEO4 days ago

Locate Anything?

Pies Avalon4 days ago

The final form of this will be an api service like jev.

Jim Wallace4 days ago

That scene in The Matrix Reloaded could use some DLSS5 treatment tho

Artzy4 days ago

The throughput on that box decoding is wild

安叫兽|Bird🕊️ 🔶 BNB4 days ago

3B 还能跑这么快，GUI 场景应该挺香，等实测了

Antonio Rocha4 days ago

É mais rapido que o YOLO 26?

Ramesh Devasi4 days ago

why i am not able to get simple 4 corner flat planer tracker which can run in web

Related Videos

NVIDIA's LocateAnything is a new vision model for grounding and detection. Very performant and accurate! > 10x faster than Qwen3-VL > 138M queries + 785M boxes > GUI, OCR, docs, dense detection > Free & open source

NVIDIA's LocateAnything is a new vision model for grounding and detection. Very performant and accurate! > 10x faster than Qwen3-VL > 138M queries + 785M boxes > GUI, OCR, docs, dense detection > Free & open source

AI Search

121,910 views • 4 months ago

Check out our demos using LFM2.5-VL-3B, our latest lightweight, vision-language model that reads screens, documents, and the physical world. First up: LFM2.5-VL-3B running fully on-device in the browser with WebGPU to understand a document page. The model parses the entire layout in one pass and returns regions and labels that the interface renders as an overlay. The demo highlights OCR and layout understanding for visually structured content such as forms, reports, receipts, and other documents. 🧵

Check out our demos using LFM2.5-VL-3B, our latest lightweight, vision-language model that reads screens, documents, and the physical world. First up: LFM2.5-VL-3B running fully on-device in the browser with WebGPU to understand a document page. The model parses the entire layout in one pass and returns regions and labels that the interface renders as an overlay. The demo highlights OCR and layout understanding for visually structured content such as forms, reports, receipts, and other documents. 🧵

Liquid AI

12,613 views • 1 month ago

Baidu Inc. just dropped PaddleOCR-VL-1.5, and it’s not just another OCR model. This is a fully open-source OCR model with just 0.9B parameters, built specifically for production-grade document understanding. And the results speak loudly 👇 On OmniDocBench V1.5, the most authoritative global document parsing benchmark, PaddleOCR-VL-1.5 ranks #1 overall, hitting 94.5% overall accuracy, outperforming models like DeepSeek-OCR2. As the first OCR model natively supporting irregular document layout positioning, it perfectly solves OCR pain points in production for complex structured business documents such as financial reports and insurance forms. While a lot of the industry is racing toward bigger models, PaddleOCR is betting on something smarter: small parameters + high performance + open source + real usability. That combination is exactly what scalable document intelligence needs right now. Demo / API → Open source → Model → Curious to see how teams start building on top of this.

Baidu Inc. just dropped PaddleOCR-VL-1.5, and it’s not just another OCR model. This is a fully open-source OCR model with just 0.9B parameters, built specifically for production-grade document understanding. And the results speak loudly 👇 On OmniDocBench V1.5, the most authoritative global document parsing benchmark, PaddleOCR-VL-1.5 ranks #1 overall, hitting 94.5% overall accuracy, outperforming models like DeepSeek-OCR2. As the first OCR model natively supporting irregular document layout positioning, it perfectly solves OCR pain points in production for complex structured business documents such as financial reports and insurance forms. While a lot of the industry is racing toward bigger models, PaddleOCR is betting on something smarter: small parameters + high performance + open source + real usability. That combination is exactly what scalable document intelligence needs right now. Demo / API → Open source → Model → Curious to see how teams start building on top of this.

Bishal Nandi

35,318 views • 8 months ago

Announcing the open-source release of Qwen3-VL! A powerful vision-language model that can operate GUIs, code charts from mockups, and recognize "everything" from daily life to specialized fields. Highlights: 🔹 Precise event location in videos up to 2 hours long. 🔹 OCR language support boosted from 19 to 32, with major gains on rare characters and tilted text. 🔹 Supports a native 256K context length, expandable to 1M tokens. 🔹 Achieves leading accuracy in risk detection in real-world scenarios. Available on ModelScope, HuggingFace, GitHub, and integrated into Alibaba Cloud Model Studio. Try it today!

Announcing the open-source release of Qwen3-VL! A powerful vision-language model that can operate GUIs, code charts from mockups, and recognize "everything" from daily life to specialized fields. Highlights: 🔹 Precise event location in videos up to 2 hours long. 🔹 OCR language support boosted from 19 to 32, with major gains on rare characters and tilted text. 🔹 Supports a native 256K context length, expandable to 1M tokens. 🔹 Achieves leading accuracy in risk detection in real-world scenarios. Available on ModelScope, HuggingFace, GitHub, and integrated into Alibaba Cloud Model Studio. Try it today!

Tongyi Lab

59,959 views • 1 year ago

LiteParse Samples 📄🧑‍🏫 LiteParse is our free and fast document parser that can parse text with bounding boxes from any document. I’ve created a new repository that contains some demos on how you can make use of its outputs! ✅ Comparison against PyPDF and PyMuPDF ✅ Visual Citations: search for any keyword and see joined bounding boxes pop up in the source images It should be a default tool to help AI agents OCR any document type (supports PDF, Word, Powerpoint + dozens of other formats), before heavier-weight VLM-based document parsing solutions (like our own LlamaParse). LiteParse Samples repo: LiteParse:

LiteParse Samples 📄🧑‍🏫 LiteParse is our free and fast document parser that can parse text with bounding boxes from any document. I’ve created a new repository that contains some demos on how you can make use of its outputs! ✅ Comparison against PyPDF and PyMuPDF ✅ Visual Citations: search for any keyword and see joined bounding boxes pop up in the source images It should be a default tool to help AI agents OCR any document type (supports PDF, Word, Powerpoint + dozens of other formats), before heavier-weight VLM-based document parsing solutions (like our own LlamaParse). LiteParse Samples repo: LiteParse:

Jerry Liu

13,196 views • 6 months ago

Speed and quality can finally coexist in diffusion-based language generation. Introducing DiDi-Instruct, a Discrete Diffusion Divergence Instruct method that distills a pre-trained discrete diffusion language model (dLLM) into a few-step student for ultra-fast generation. Built on integral KL-divergence minimization, DiDi-Instruct achieves up to 64× faster decoding, surpasses both its teacher and GPT-2, and cuts training time by 20×. Ultra-Fast Language Generation via Discrete Diffusion Divergence Instruct Paper: Code: Project: Our report: 📬 #PapersAccepted by Jiqizhixin

Speed and quality can finally coexist in diffusion-based language generation. Introducing DiDi-Instruct, a Discrete Diffusion Divergence Instruct method that distills a pre-trained discrete diffusion language model (dLLM) into a few-step student for ultra-fast generation. Built on integral KL-divergence minimization, DiDi-Instruct achieves up to 64× faster decoding, surpasses both its teacher and GPT-2, and cuts training time by 20×. Ultra-Fast Language Generation via Discrete Diffusion Divergence Instruct Paper: Code: Project: Our report: 📬 #PapersAccepted by Jiqizhixin

机器之心 JIQIZHIXIN

18,196 views • 11 months ago

🤯 Building computer-vision applications is never going to be the same again. As Andrej Karpathy says, "English is going to be the new programming-language", and we just made that happen for programming visual AI. 👨‍💻 This past month, the VLM Run team cooked up something magical that I'm especially proud of - a natural-language interface to build and run computer vision workflows (object-detection, OCR, parsing, and so much more) – all via our new VLM Run MCP server. 🔌 It works with any MCP-compatible AI agent (Claude, OpenAI, Cursor, etc.) and abstracts complex visual workflows into simple English commands. No need to write Python or fiddle with bespoke computer-vision code – just say what you want your agent to do, and it does it. Here’s a glimpse into what you can build today ( ☑️ Image AI agents – run detection, tagging, blurring, cropping, and more traditional CV tasks with English ☑️ Document AI agents – extract totals from invoices, search visual content, or act on what’s found ☑️ Video AI agents – understand scene content, trim clips, add subtitles, and create ready-to-share edits ☑️ And best-of-all, it's fully programmable with English. 🧑‍🍳 MCP Showcase: 📚 MCP Cookbooks: 📚 MCP Docs: #computervision #ai #agents #llm #vlm #mcp #vibecoding #genai

🤯 Building computer-vision applications is never going to be the same again. As Andrej Karpathy says, "English is going to be the new programming-language", and we just made that happen for programming visual AI. 👨‍💻 This past month, the VLM Run team cooked up something magical that I'm especially proud of - a natural-language interface to build and run computer vision workflows (object-detection, OCR, parsing, and so much more) – all via our new VLM Run MCP server. 🔌 It works with any MCP-compatible AI agent (Claude, OpenAI, Cursor, etc.) and abstracts complex visual workflows into simple English commands. No need to write Python or fiddle with bespoke computer-vision code – just say what you want your agent to do, and it does it. Here’s a glimpse into what you can build today ( ☑️ Image AI agents – run detection, tagging, blurring, cropping, and more traditional CV tasks with English ☑️ Document AI agents – extract totals from invoices, search visual content, or act on what’s found ☑️ Video AI agents – understand scene content, trim clips, add subtitles, and create ready-to-share edits ☑️ And best-of-all, it's fully programmable with English. 🧑‍🍳 MCP Showcase: 📚 MCP Cookbooks: 📚 MCP Docs: #computervision #ai #agents #llm #vlm #mcp #vibecoding #genai

Sudeep Pillai

17,789 views • 1 year ago

Open Sourcing Forge: 3D Gaussian splat rendering for web developers! 3DGS has become a dominant paradigm for differentiable rendering, combining high visual quality and real-time rendering. However, support for splatting on the web still lags behind its adoption in AI.

Open Sourcing Forge: 3D Gaussian splat rendering for web developers! 3DGS has become a dominant paradigm for differentiable rendering, combining high visual quality and real-time rendering. However, support for splatting on the web still lags behind its adoption in AI.

spark

120,072 views • 1 year ago

If you’re building an “agent over your PDFs” in 2026, one of the most underrated things you can do is to provide exact grounding back to the source document for any agentic answers. Specifically, showing the exact word, line, or region that the answer came from. Frontier vision models do a terrible job at this. They have a hard time predicting exact bounding boxes over region, tying it back to the source answer, and surfacing it as context in a cost-effective and accurate manner. Dedicated VLM-based document OCR/extraction tools like LlamaParse are a lot better at visual grounding and specifically tying it to extracted outputs. One of our main evaluation areas in ExtractBench, our recently released real-world benchmark, is to evaluate the visual grounding capabilities of document extraction. ExtractBench: If you want to give LlamaParse a shot, come check it out:

If you’re building an “agent over your PDFs” in 2026, one of the most underrated things you can do is to provide exact grounding back to the source document for any agentic answers. Specifically, showing the exact word, line, or region that the answer came from. Frontier vision models do a terrible job at this. They have a hard time predicting exact bounding boxes over region, tying it back to the source answer, and surfacing it as context in a cost-effective and accurate manner. Dedicated VLM-based document OCR/extraction tools like LlamaParse are a lot better at visual grounding and specifically tying it to extracted outputs. One of our main evaluation areas in ExtractBench, our recently released real-world benchmark, is to evaluate the visual grounding capabilities of document extraction. ExtractBench: If you want to give LlamaParse a shot, come check it out:

Jerry Liu

14,515 views • 1 month ago

A toolkit for building agents that watch, listen, and understand video. Low latency by design. Open source. Production ready. Vision Agents lets you build real time video AI that works with your models and your edge layer. Supports YOLO, Moondream, Cartesia, Deepgram, ElevenLabs, HeyGen, Gemini, OpenAI, and more. Quick model switching. Easy to use API. Perfect for coaching tools, collaboration apps, avatars, and robotics.

A toolkit for building agents that watch, listen, and understand video. Low latency by design. Open source. Production ready. Vision Agents lets you build real time video AI that works with your models and your edge layer. Supports YOLO, Moondream, Cartesia, Deepgram, ElevenLabs, HeyGen, Gemini, OpenAI, and more. Quick model switching. Easy to use API. Perfect for coaching tools, collaboration apps, avatars, and robotics.

GitHub Projects Community

30,274 views • 9 months ago

Introducing VL-JEPA: Vision-Language Joint Embedding Predictive Architecture for streaming, live action recognition, retrieval, VQA, and classification tasks with better performance and higher efficiency than large VLMs. • VL-JEPA is the first non-generative model that can perform general-domain vision-language tasks in real-time, built on a joint embedding predictive architecture. • We demonstrate in controlled experiments that VL-JEPA, trained with latent space embedding prediction, outperforms VLMs that rely on data space token prediction. • We show that VL-JEPA delivers significant efficiency gains over VLMs for online video streaming applications, thanks to its non-autoregressive design and native support for selective decoding. • We highlight that our VL-JEPA model, with an unified model architecture, can effectively handle a wide range of classification, retrieval, and VQA tasks at the same time. by Delong Chen (陈德龙) Mustafa Shukor Théo Moutakanni Willy Jade Lei Yu Tejaswi Kasarla Allen Bolourchi Yann LeCun Pascale Fung

Introducing VL-JEPA: Vision-Language Joint Embedding Predictive Architecture for streaming, live action recognition, retrieval, VQA, and classification tasks with better performance and higher efficiency than large VLMs. • VL-JEPA is the first non-generative model that can perform general-domain vision-language tasks in real-time, built on a joint embedding predictive architecture. • We demonstrate in controlled experiments that VL-JEPA, trained with latent space embedding prediction, outperforms VLMs that rely on data space token prediction. • We show that VL-JEPA delivers significant efficiency gains over VLMs for online video streaming applications, thanks to its non-autoregressive design and native support for selective decoding. • We highlight that our VL-JEPA model, with an unified model architecture, can effectively handle a wide range of classification, retrieval, and VQA tasks at the same time. by Delong Chen (陈德龙) Mustafa Shukor Théo Moutakanni Willy Jade Lei Yu Tejaswi Kasarla Allen Bolourchi Yann LeCun Pascale Fung

Pascale Fung

90,144 views • 9 months ago

Okay so Jev can actually do computer use really well Without any screenshots, or LLMs and no Pixels leave my mac I dont even read the Dom elements A local CoreML model segments every button and UI element on screen. On-device OCR reads the labels. That text is all Jev gets. It returns a probability across those elements and tells me the best one to click. Then it clicks, re-runs detection, and decides again. In a loop until the goal is done. ~90ms per decision. Faster than any LLM computer use I've tried. Blazing fast computer use, without any latency TypeSafe AI is building something really interesting

Okay so Jev can actually do computer use really well Without any screenshots, or LLMs and no Pixels leave my mac I dont even read the Dom elements A local CoreML model segments every button and UI element on screen. On-device OCR reads the labels. That text is all Jev gets. It returns a probability across those elements and tells me the best one to click. Then it clicks, re-runs detection, and decides again. In a loop until the goal is done. ~90ms per decision. Faster than any LLM computer use I've tried. Blazing fast computer use, without any latency TypeSafe AI is building something really interesting

Milind S

241,887 views • 16 days ago

Introducing Phoenix Deep Research, a vertical AI agent for in-depth, multi-step research and real-time knowledge. Phoenix Deep Research is blazingly fast with average response output time of 30 to 90 seconds, much faster than industry averages and is built on top of an accelerated Deepseek R1 utilizing SkyNet High Performance Cluster consisting of the newest Huawei Ascend and Nvidia A100, and A10 GPUs. Phoenix DR combines the most cutting-edge open source LLM, industry-specific agentic workflows, real-time AI search, and novel decentralized compute infrastructure into a differentiated and versatile tool previously unseen in the decentralized AI space. Phoenix DR will be open for Beta whitelist and testing early next week for community users and partners. Stay tuned. #Phoenix $PHB #DeepResearch

Introducing Phoenix Deep Research, a vertical AI agent for in-depth, multi-step research and real-time knowledge. Phoenix Deep Research is blazingly fast with average response output time of 30 to 90 seconds, much faster than industry averages and is built on top of an accelerated Deepseek R1 utilizing SkyNet High Performance Cluster consisting of the newest Huawei Ascend and Nvidia A100, and A10 GPUs. Phoenix DR combines the most cutting-edge open source LLM, industry-specific agentic workflows, real-time AI search, and novel decentralized compute infrastructure into a differentiated and versatile tool previously unseen in the decentralized AI space. Phoenix DR will be open for Beta whitelist and testing early next week for community users and partners. Stay tuned. #Phoenix $PHB #DeepResearch

Phoenix AI Labs

21,303 views • 1 year ago

A big milestone for AI on Windows: gpt‑oss‑20b is now available in Windows AI Foundry and Azure AI Foundry! You can now run this open-source, frontier-class model locally on a high-performance Windows PC. It’s fast, capable, and built for real-world use. Brought to life through collaboration across Windows, Azure, and OpenAI teams. We can’t wait to see what you create. 🚀

A big milestone for AI on Windows: gpt‑oss‑20b is now available in Windows AI Foundry and Azure AI Foundry! You can now run this open-source, frontier-class model locally on a high-performance Windows PC. It’s fast, capable, and built for real-world use. Brought to life through collaboration across Windows, Azure, and OpenAI teams. We can’t wait to see what you create. 🚀

Pavan Davuluri

80,597 views • 1 year ago

New LLMs that control UIs! ByteDance Research releases UI-TARS, fine-tuned GUI agent that integrates reasoning, and action capabilities into a single vision-language model. Think of computer use but open. 👀 TL;DR; 3️⃣ Available in 3 sizes: 2B, 7B, and 72B parameters 🧠 Trained Qwen2-VL models with SFT & DPO 🥇 72B version achieves 82.8% on VisualWebBench (beating GPT-4 and Claude) 🏆 Achieves state-of-the-art results on 10+ GUI agent benchmarks 💡 Reasons before taking an action 🧑🏻‍💻 Can Click, Long Press, type, scroll, open app, navigate back/home, wait 🤗 Released under Apache 2.0 on Hugging Face

New LLMs that control UIs! ByteDance Research releases UI-TARS, fine-tuned GUI agent that integrates reasoning, and action capabilities into a single vision-language model. Think of computer use but open. 👀 TL;DR; 3️⃣ Available in 3 sizes: 2B, 7B, and 72B parameters 🧠 Trained Qwen2-VL models with SFT & DPO 🥇 72B version achieves 82.8% on VisualWebBench (beating GPT-4 and Claude) 🏆 Achieves state-of-the-art results on 10+ GUI agent benchmarks 💡 Reasons before taking an action 🧑🏻‍💻 Can Click, Long Press, type, scroll, open app, navigate back/home, wait 🤗 Released under Apache 2.0 on Hugging Face

Philipp Schmid

48,170 views • 1 year ago

People are overlooking Google Gemini Realtime models for computer use It gave me sub 100ms latency with computer use. It has a much larger context window and is much cheaper as well Combine that with local OCR and local screen detection model based on Omniparser by Microsoft it works under 100ms action taking when combined with Cua I also put in a harness for Nous Research Hermes with it. You can access it all at your tip of your cursor. You can draw on your screen to give your agents a context And I am making it Open Source! Link in the Comments Sundar Pichai Min-Liang Tan Mojtaba Seyedhosseini

People are overlooking Google Gemini Realtime models for computer use It gave me sub 100ms latency with computer use. It has a much larger context window and is much cheaper as well Combine that with local OCR and local screen detection model based on Omniparser by Microsoft it works under 100ms action taking when combined with Cua I also put in a harness for Nous Research Hermes with it. You can access it all at your tip of your cursor. You can draw on your screen to give your agents a context And I am making it Open Source! Link in the Comments Sundar Pichai Min-Liang Tan Mojtaba Seyedhosseini

Milind S

38,361 views • 4 months ago

Smart Turn v2: open source, native audio turn detection in 14 languages. New checkpoint of the open source, open data, open training code, semantic VAD model on Hugging Face, fal, and Pipecat AI. - 3x faster inference (12ms on an L40) - 14 languages (13 more than v1, which was english-only) - New synthetic data set `chirp_3_all` with ~163k audio samples - 99% accuracy on held out `human_5_all` test data Good turn detection is critical for voice agents. This model "understands" both semantic and audio patterns, and mitigates the voice AI trade-off between unwanted turn latency vs the agent interrupting people before they are finished speaking. Training scripts for both Modal and local training are in the repo. We want to make it as easy as possible to contribute to or customize this model! Here's a demo running the smart-turn model with default settings, aimed at generally hitting 400ms total turn detection time. You can tune things to be faster, too. You can help by contributing data, doing architecture expermints, or cleaning open source data! Keep reading ...

Smart Turn v2: open source, native audio turn detection in 14 languages. New checkpoint of the open source, open data, open training code, semantic VAD model on Hugging Face, fal, and Pipecat AI. - 3x faster inference (12ms on an L40) - 14 languages (13 more than v1, which was english-only) - New synthetic data set `chirp_3_all` with ~163k audio samples - 99% accuracy on held out `human_5_all` test data Good turn detection is critical for voice agents. This model "understands" both semantic and audio patterns, and mitigates the voice AI trade-off between unwanted turn latency vs the agent interrupting people before they are finished speaking. Training scripts for both Modal and local training are in the repo. We want to make it as easy as possible to contribute to or customize this model! Here's a demo running the smart-turn model with default settings, aimed at generally hitting 400ms total turn detection time. You can tune things to be faster, too. You can help by contributing data, doing architecture expermints, or cleaning open source data! Keep reading ...

kwindla

42,246 views • 1 year ago

LiteParse is the best model-free, open-source document parser for AI agents. It now gets a first-class landing page on our website 💫 Our company mission is building the world's best agentic document processing platform, and liteparse is the central pillar behind our OSS efforts. It's blazing fast (and getting faster soon!), supports 50+ file formats, and is one-shot installable as an agent skill. Webpage: Come check it out:

LiteParse is the best model-free, open-source document parser for AI agents. It now gets a first-class landing page on our website 💫 Our company mission is building the world's best agentic document processing platform, and liteparse is the central pillar behind our OSS efforts. It's blazing fast (and getting faster soon!), supports 50+ file formats, and is one-shot installable as an agent skill. Webpage: Come check it out:

Jerry Liu

70,895 views • 5 months ago