We release the most comprehensive 𝗲𝗺𝗽𝗶𝗿𝗶𝗰𝗮𝗹 𝘀𝘁𝘂𝗱𝘆 𝗼𝗳 𝗚𝗣𝗧-𝟲-𝗔𝘀𝘁𝗿𝗮’𝘀 𝗲𝗺𝗯𝗼𝗱𝗶𝗲𝗱 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 across 12 simulation benchmarks and 8 real-robot platforms, together with 𝗦𝗲𝗻𝘀𝗲𝗡𝗼𝘃𝗮-𝗥𝗼𝗯𝗼𝗥𝗦𝗜: an open-source harness framework and a 27B model that runs at ~1/50 the cost of GPT-6-Astra. Here's what we found: 🔹 𝗧𝗵𝗲 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗰𝗲𝗶𝗹𝗶𝗻𝗴 𝗶𝘀 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗺𝘂𝗰𝗵 𝗵𝗶𝗴𝗵𝗲𝗿. Under a unified protocol (ICL + up to 10× the interaction budget), GPT-6-Astra achieves clear SOTA across all 12 benchmarks: RoboDojo: 27.6% at 1× → 81.9% at 10×, DexJoCo: 87.3% vs. 40.3% prior SOTA, while the LIBERO benchmarks are pushed to saturation. 🔹 𝗧𝗼𝗼 𝗺𝘂𝗰𝗵 𝗯𝘂𝗰𝗸 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗯𝗮𝗻𝗴. Even a simple task like placing pens in a holder costs ~$55 in inference. 🔹 𝗧𝗵𝗲 𝘄𝗼𝗿𝗹𝗱 𝗱𝗼𝗲𝘀𝗻’𝘁 𝘄𝗮𝗶𝘁. Time matters as real robots overheat, and an egg burned while the model was still thinking. Our takeaway: GPT-6-Astra has largely crossed the threshold of 𝗲𝗺𝗯𝗼𝗱𝗶𝗲𝗱 𝗰𝗼𝗺𝗽𝗲𝘁𝗲𝗻𝗰𝗲, but 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗮𝗹 𝗲𝗺𝗯𝗼𝗱𝗶𝗲𝗱 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 is not here yet. 🌐 Project page: sensenova.ai/has-gpt-6-astra… 📄 Report: sensenova.ai/gpt-6/assets/do… 💻 Code & model: github.com/OpenSenseNova/Sen…

𝗔 𝗗𝗼𝘇𝗲𝗻 𝗬𝗲𝗮𝗿𝘀. 𝗔 𝗣𝗮𝘀𝘀𝗶𝗼𝗻 𝗧𝗵𝗮𝘁 𝗕𝘂𝗿𝗻𝘀 𝗢𝗻. At our desks, on stage, and on the field, 𝘄𝗲 𝗴𝗶𝘃𝗲 𝗶𝘁 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴 𝘄𝗲'𝘃𝗲 𝗴𝗼𝘁. Going all in is how we truly enjoy what we do—and make it our own. Twelve years in, 𝘄𝗲 𝗹𝗶𝘃𝗲 𝗯𝘆 𝗼𝘂𝗿 “𝗯𝗹𝗮𝗰𝗸 𝘀𝗵𝗲𝗲𝗽” 𝘀𝗽𝗶𝗿𝗶𝘁: 𝗱𝗮𝗿𝗶𝗻𝗴 𝘁𝗼 𝗶𝗻𝗻𝗼𝘃𝗮𝘁𝗲, 𝗱𝗮𝗿𝗶𝗻𝗴 𝘁𝗼 𝗯𝗲 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁. 𝗗𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁, 𝗯𝘂𝘁 𝗻𝗲𝘃𝗲𝗿 𝗮𝗹𝗼𝗻𝗲. 🔥 #12Anniversary

“𝗙𝗼𝗿 𝗺𝗼𝘀𝘁 𝗽𝗲𝗼𝗽𝗹𝗲, 𝗔𝗜 𝗶𝘀 𝘁𝗵𝗲 𝗶𝗰𝗶𝗻𝗴 𝗼𝗻 𝘁𝗵𝗲 𝗰𝗮𝗸𝗲. 𝗙𝗼𝗿 𝗽𝗲𝗼𝗽𝗹𝗲 𝘄𝗶𝘁𝗵 𝗱𝗶𝘀𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀, 𝗶𝘁'𝘀 𝗹𝗶𝗸𝗲 𝗳𝗶𝗻𝗱𝗶𝗻𝗴 𝗮 𝗳𝗶𝗿𝗲 𝗶𝗻 𝘁𝗵𝗲 𝗳𝗿𝗲𝗲𝘇𝗶𝗻𝗴 𝗰𝗼𝗹𝗱,” 𝘀𝗮𝘆𝘀 𝗖𝗮𝗼 𝗪𝗲𝗻𝘄𝗲𝗶, 𝗮 𝘃𝗶𝘀𝘂𝗮𝗹𝗹𝘆 𝗶𝗺𝗽𝗮𝗶𝗿𝗲𝗱 𝗮𝗱𝘃𝗼𝗰𝗮𝘁𝗲 𝗳𝗼𝗿 𝗮𝗰𝗰𝗲𝘀𝘀𝗶𝗯𝗹𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗶𝗻 𝗖𝗵𝗶𝗻𝗮. Observed today, 𝟭𝟱 𝗢𝗰𝘁𝗼𝗯𝗲𝗿, 𝗜𝗻𝘁𝗲𝗿𝗻𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗪𝗵𝗶𝘁𝗲 𝗖𝗮𝗻𝗲 𝗦𝗮𝗳𝗲𝘁𝘆 𝗗𝗮𝘆 highlights the importance of accessibility and equal participation for people with visual impairments. Alongside running a massage shop, Cao converts textbooks and learning materials for students with visual and multiple disabilities. It's no easy task: some books run to a thousand pages, and some have no digital or Braille version at all. Cao relies on a screen reader, software that reads text aloud, bit by bit. This is where 𝗦𝗲𝗻𝘀𝗲𝗧𝗶𝗺𝗲 𝗥𝗮𝗰𝗰𝗼𝗼𝗻 𝗪𝗼𝗿𝗸, 𝗼𝘂𝗿 𝗔𝗜 𝗼𝗳𝗳𝗶𝗰𝗲 𝗮𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁, comes in. 𝗜𝘁 𝗵𝗲𝗹𝗽𝘀 𝗵𝗶𝗺 𝗽𝗿𝗼𝗰𝗲𝘀𝘀 𝗮𝗻𝗱 𝗼𝗿𝗴𝗮𝗻𝗶𝘀𝗲 𝗹𝗲𝗻𝗴𝘁𝗵𝘆 𝗱𝗼𝗰𝘂𝗺𝗲𝗻𝘁𝘀 𝗰𝗼𝗻𝘁𝗮𝗶𝗻𝗶𝗻𝗴 𝘁𝗲𝘅𝘁, 𝗶𝗺𝗮𝗴𝗲𝘀 𝗮𝗻𝗱 𝘁𝗮𝗯𝗹𝗲𝘀, 𝗺𝗮𝗸𝗶𝗻𝗴 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗺𝗮𝘁𝗲𝗿𝗶𝗮𝗹𝘀 𝗲𝗮𝘀𝗶𝗲𝗿 𝘁𝗼 𝗻𝗮𝘃𝗶𝗴𝗮𝘁𝗲 𝘄𝗶𝘁𝗵 𝗮 𝘀𝗰𝗿𝗲𝗲𝗻 𝗿𝗲𝗮𝗱𝗲𝗿. Cao wanted to do more. 𝗪𝗼𝗿𝗸𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗥𝗮𝗰𝗰𝗼𝗼𝗻 𝗪𝗼𝗿𝗸 𝘁𝗼 𝘁𝘂𝗿𝗻 𝗵𝗶𝘀 𝗶𝗱𝗲𝗮 𝗶𝗻𝘁𝗼 𝗮 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗮𝗹 𝘀𝗼𝗹𝘂𝘁𝗶𝗼𝗻, 𝗵𝗲 𝗯𝘂𝗶𝗹𝘁 “𝗟𝗶𝗴𝗵𝘁 𝗕𝗿𝗶𝗱𝗴𝗲 𝗔𝗜,” 𝗮 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗱𝗲𝘀𝗶𝗴𝗻𝗲𝗱 𝘁𝗼 𝗵𝗲𝗹𝗽 𝘁𝗲𝗮𝗰𝗵𝗲𝗿𝘀, 𝗽𝗮𝗿𝗲𝗻𝘁𝘀 𝗮𝗻𝗱 𝘃𝗼𝗹𝘂𝗻𝘁𝗲𝗲𝗿𝘀 𝘀𝘂𝗽𝗽𝗼𝗿𝘁 𝗰𝗵𝗶𝗹𝗱𝗿𝗲𝗻 𝘄𝗶𝘁𝗵 𝘃𝗶𝘀𝘂𝗮𝗹 𝗮𝗻𝗱 𝗺𝘂𝗹𝘁𝗶𝗽𝗹𝗲 𝗱𝗶𝘀𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀. 𝗜𝘁 𝗮𝗻𝗮𝗹𝘆𝘀𝗲𝘀 𝗲𝗮𝗰𝗵 𝘀𝘁𝘂𝗱𝗲𝗻𝘁'𝘀 𝗰𝗶𝗿𝗰𝘂𝗺𝘀𝘁𝗮𝗻𝗰𝗲𝘀 𝗮𝗻𝗱 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲𝘀 𝘁𝗮𝗶𝗹𝗼𝗿𝗲𝗱 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗺𝗮𝘁𝗲𝗿𝗶𝗮𝗹𝘀, 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗽𝗹𝗮𝗻𝘀 𝗮𝗻𝗱 𝘁𝗲𝗮𝗰𝗵𝗶𝗻𝗴 𝗽𝗹𝗮𝗻𝘀, 𝗮𝗹𝗼𝗻𝗴 𝘄𝗶𝘁𝗵 𝗿𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝘀 𝗳𝗼𝗿 𝘁𝗲𝗮𝗰𝗵𝗲𝗿𝘀 𝗮𝗻𝗱 𝗽𝗮𝗿𝗲𝗻𝘁𝘀. Raccoon Work also helps Cao review his own PowerPoint slides. 𝗨𝘀𝗶𝗻𝗴 𝗶𝘁𝘀 𝗶𝗺𝗮𝗴𝗲-𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀, 𝗵𝗲 𝗰𝗮𝗻 𝗴𝗲𝘁 𝗱𝗲𝘁𝗮𝗶𝗹𝗲𝗱 𝗱𝗲𝘀𝗰𝗿𝗶𝗽𝘁𝗶𝗼𝗻𝘀 𝗼𝗳 𝘀𝗹𝗶𝗱𝗲 𝗰𝗼𝗻𝘁𝗲𝗻𝘁 𝗮𝗻𝗱 𝗹𝗮𝘆𝗼𝘂𝘁, 𝗮𝗹𝗹𝗼𝘄𝗶𝗻𝗴 𝗵𝗶𝗺 𝘁𝗼 𝗿𝗲𝘃𝗶𝗲𝘄 𝗮𝗻𝗱 𝗿𝗲𝗳𝗶𝗻𝗲 𝗽𝗿𝗲𝘀𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻𝘀 𝗶𝗻𝗱𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝘁𝗹𝘆. “Light Bridge AI” won Silver in the finals of the SenseTime Raccoon Work OPC (One Person Company) Challenge. Accessibility benefits us all. Age, illness or injury can affect anyone's ability to access information. Making information more accessible can create a better world for everyone. 𝗪𝗲 𝗵𝗼𝗽𝗲 𝘁𝗵𝗮𝘁 𝗼𝗻𝗲 𝗱𝗮𝘆, 𝘃𝗶𝘀𝘂𝗮𝗹𝗹𝘆 𝗶𝗺𝗽𝗮𝗶𝗿𝗲𝗱 𝗰𝗵𝗶𝗹𝗱𝗿𝗲𝗻 𝘄𝗶𝗹𝗹 𝗻𝗼 𝗹𝗼𝗻𝗴𝗲𝗿 𝗵𝗲𝗮𝗿, “𝗬𝗼𝘂 𝗰𝗮𝗻'𝘁 𝗮𝗰𝗰𝗲𝘀𝘀 𝘁𝗵𝗶𝘀,” 𝗯𝘂𝘁 𝗶𝗻𝘀𝘁𝗲𝗮𝗱, “𝗟𝗲𝘁'𝘀 𝗳𝗶𝗻𝗱 𝗮𝗻𝗼𝘁𝗵𝗲𝗿 𝘄𝗮𝘆.” #InternationalWhiteCaneSafetyDay #Accessibility #AIforGood #RaccoonWork

感谢关注！追踪我们，获取更多 AI 前沿资讯🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Thank you. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

🤖 𝗘𝘃𝗲𝗿𝘆 𝗿𝗼𝗯𝗼𝘁 𝘁𝗮𝘀𝗸 𝗳𝗮𝗶𝗹𝘂𝗿𝗲 𝗰𝗮𝗻 𝗯𝗲𝗰𝗼𝗺𝗲 𝗮 𝗹𝗲𝘀𝘀𝗼𝗻 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗻𝗲𝘅𝘁 𝗶𝘁𝗲𝗿𝗮𝘁𝗶𝗼𝗻. 𝗦𝗲𝗻𝘀𝗲𝗡𝗼𝘃𝗮-𝗥𝗼𝗯𝗼𝗥𝗦𝗜 uses 𝗥𝗦𝗜 to evolve 𝗮𝗴𝗲𝗻𝘁 𝗵𝗮𝗿𝗻𝗲𝘀𝘀𝗲𝘀 through physical task feedback, without updating model parameters. ▶ 𝗥𝗼𝗯𝗼𝗗𝗼𝗷𝗼: 56.83 avg. score, 50.83% success rate, +𝟮𝟳.𝟴𝟲 𝗽𝗼𝗶𝗻𝘁𝘀 𝘃𝘀. 𝘁𝗵𝗲 𝗚𝗣𝗧-𝟲 𝗔𝘀𝘁𝗿𝗮 𝗯𝗮𝘀𝗲𝗹𝗶𝗻𝗲 (𝟮𝟴.𝟵𝟳) ▶ 𝗥𝗦𝗜 𝗹𝗼𝗼𝗽: Execute → Diagnose → Improve → Validate → Inherit ▶ 𝗘𝘅𝗽𝗹𝗼𝗿𝗲𝗱 𝗮𝗻𝗱 𝘃𝗮𝗹𝗶𝗱𝗮𝘁𝗲𝗱: Multi-point EEF prediction and Planning + Feedback Subagents This achievement reflects the collective efforts of our joint team. RoboRSI will be 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲𝗱 𝘀𝗼𝗼𝗻 at github.com/OpenSenseNova/Sen…. A comprehensive 𝘁𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗿𝗲𝗽𝗼𝗿𝘁 detailing our methods and findings is also on the way. Stay tuned!

Thank you. We have sent you a DM. Follow us for updates on AI🙌: xitter.cf/SenseTime_AI

Is it time to clock out yet? 🫠

Busy as a bee, literally. 🐝 The garden system map is in motion.

A new angle on creativity. 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝗗𝗲𝘀𝗶𝗴𝗻, 𝗽𝗼𝘄𝗲𝗿𝗲𝗱 𝗯𝘆 𝗦𝗲𝗻𝘀𝗲𝗡𝗼𝘃𝗮 𝟲.𝟴 𝗙𝗹𝗮𝘀𝗵 𝗣𝗿𝗲𝘃𝗶𝗲𝘄, 𝗯𝗿𝗶𝗻𝗴𝘀 𝘀𝘁𝗮𝘁𝗶𝗰 𝗶𝗺𝗮𝗴𝗲𝘀 𝘁𝗼 𝗹𝗶𝗳𝗲.

Follow us for updates on AI:🤲 xitter.cf/SenseTime_AI

He never sits still. Neither does our poster. #SunWukong

Sometimes, you just have to find your own way out. 🐧 🎬 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝗗𝗲𝘀𝗶𝗴𝗻, 𝗽𝗼𝘄𝗲𝗿𝗲𝗱 𝗯𝘆 𝗦𝗲𝗻𝘀𝗲𝗡𝗼𝘃𝗮 𝟲.𝟴 𝗙𝗹𝗮𝘀𝗵 𝗣𝗿𝗲𝘃𝗶𝗲𝘄, 𝗯𝗿𝗶𝗻𝗴𝘀 𝘀𝘁𝗮𝘁𝗶𝗰 𝗶𝗺𝗮𝗴𝗲𝘀 𝘁𝗼 𝗹𝗶𝗳𝗲.