模型发布VIDEO
Odyssey 发布世界模型 Odyssey-3 研究预览
Odyssey-3 可实时生成可交互环境,物体能随用户操作移动反应,已开放研究预览,并测试用于机器人与自驾训练。
The Daily Block
@thedailyblock24 天前收录原推发布于 2026.10.09
The Daily Block @thedailyblock原推发布于 2026.10.09
🚨 ODYSSEY LAUNCHES ODYSSEY-3, ITS MOST POWERFUL AI WORLD MODEL YET Odyssey-3 can create interactive environments in real time, where objects move, react and respond to what users do. The model is also being tested for robotics, self-driving systems and AI agent training. Odyssey reports that its Pro model achieved a score of 66.1 on the Physics-IQ video-to-video benchmark using best-of-8 sampling. The research preview is now available to try : https://experience.odyssey.systems/
这条是引用着另一条写的:x.com/odysseyml/status/2108238378751967385
x.com/thedailyblock/status/2108244862327771390采集于 2026.10.09 11:14