Microsoft Research· Akshay Nambi, Yash Pandya, Sahil Gupta, Sarthak Harne, Kavyansh Chourasia, Yash Lara, Ahmed Awadallah·· 2026-07-31
Echoverse: Deep, evolving environments for computer-use agents
Echoverse: Deep, evolving environments for computer-use agents
AI summary
Microsoft Research proposed Echoverse, constructing twelve training worlds for computer-use agents: ten deep domain worlds and two capability worlds. Code, data and scorers for four worlds are open-sourced.
Selection record
Threshold 60Official, first-handFirst 71Second 62
AdmittedSum of both 133 ≥ twice the threshold 120
- Source tier
- Official, first-hand; this tier's threshold is 60
- Pre-filter
- passed：微软研究发布计算机使用智能体训练环境
- Why it was chosen
- The construction process and training results show how high-fidelity synthetic environments influence capability transfer in computer-use agents.
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Microsoft Research · microsoft.com