✨ 𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗩𝗟𝗠 𝗥𝘂𝗻 𝗢𝗿𝗶𝗼𝗻 – our new visual agent that 𝘴𝘦𝘦𝘴, 𝘳𝘦𝘢𝘴𝘰𝘯𝘴, 𝘢𝘯𝘥 𝘢𝘤𝘵𝘴. Today’s frontier Vision-Language Models (VLMs) – GPT-5, Claude 4.5, Gemini 2.5 Pro – can describe what they see … But they can’t 𝗮𝗰𝘁 on it. 👉

RIP #sreenivaasan. Malayalam cinema just won’t be the same without you. You made us laugh and cry, sometimes in the same movie, spanning multiple decades. A true legend.

No one will replace him🥹

I wanted to share the (quick) story on what led us to creating Orion. And as every great product story goes, it was because of our customers. Every customer call followed the same pattern. They'd show up with a specific computer vision problem, we'd know exactly how to solve it

If you’re at #NeurIPS2025 this year and are looking to work on bleeding-edge VLMs, we’re hiring for research, engineer, and devrel roles at @vlmrun. Chat with @dineshredy in person or send us an email at hiring@vlm.run. https://t.co/rGR0IDZ2x1

📍 Live from #NeurIPS! We're looking for the best minds to join us at @vlmrun. If you're passionate about building the future of Vision Language Agents, come say hi or slide into our DMs. We are hiring: https://t.co/KrDsDTd6jP #AI #MachineLearning #Hiring #VLM #GenAI

🚀 Announcing the Orion Agent API – our new visual agent that 𝘴𝘦𝘦𝘴, 𝘳𝘦𝘢𝘴𝘰𝘯𝘴, 𝘢𝘯𝘥 𝘢𝘤𝘵𝘴. One unified chat completions API for all visual modalities: images, video, documents, and even 3D. API Docs: https://t.co/DdR22QN7za

Happy Thanksgiving from @vlmrun!

📣 Learn more about Orion: https://t.co/Ckj6sGoiCR 💬 Chat with Orion:

It's been exactly 14 days since we launched our new visual agent, Orion. With the influx of positive responses combined with the holiday season, it's made me even more grateful for my team. For those who don't know, we actually spent longer than planned getting our Orion ready.

😂😂

when you ask a PhD student when they’ll finish

📣 Learn more about Orion: https://t.co/Ckj6sGoiCR 💬 Chat with Orion:

Two weeks ago, we launched our new visual agent, Orion. Many have since asked why we chose that name. Here's the quick story: Most people know Orion's Belt. But very few can name the individual stars that form it. All you see is one unified pattern – a whole greater than its

🤓 Stoked to finally publish the Orion technical whitepaper today - it's a real banger! If you're a CV researcher or a developer building in visual AI, we'd love for you to try it out and give us feedback. We'll be opening up API access soon! 📕 Whitepaper:

Source:

Wow @GoogleDeepMind Gemini 3 really cooked 🤯 ! Look at those step improvements in Humanity's Last Exam, ARC-AGI-2, MathArena Apex, ScreenSpot-Pro and SimpleQA Verified #gemini

cc @Smol_AI @_akhaliq @mervenoyann

@vlmrun Orion marks the shift from passive visual understanding to autonomous visual reasoning. Learn more → https://t.co/W95Ko5uQOk Try it now → https://t.co/n2Oc1xrlZ2 Turn pixels into possibilities. #Orion #VLMRun #GenAI #AgenticAI #Launch

@vlmrun What Orion can do today 👇 - 𝗗𝗲𝘁𝗲𝗰𝘁: objects, faces, people - 𝗦𝗲𝗴𝗺𝗲𝗻𝘁: isolate regions interactively - 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲: edit, remix, re-imagine visuals - 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺 𝘃𝗶𝗱𝗲𝗼𝘀: trim, sample, create highlights - 𝗣𝗮𝗿𝘀𝗲 𝗱𝗼𝗰𝘂𝗺𝗲𝗻𝘁𝘀: OCR,

At @vlmrun, we re-imagined what visual intelligence should look like from the ground up. ✨ 𝗢𝗿𝗶𝗼𝗻 is the first visual agent that can 𝘴𝘦𝘦, 𝘳𝘦𝘢𝘴𝘰𝘯, 𝘢𝘯𝘥 𝘢𝘤𝘵 across any visual input – image, video, or document – inside a unified chat-completions interface. No