Offloop D1多智能体协调层:3B参数,路由基准SOTA
Offloop搞了个D1路由层,3B参数就能自动调度多个AI agent,基准上干掉了Claude Code和Codex,挺有意思。
Offloop团队推出D1层,一个仅3B参数的多智能体路由层。它在GDPval基准上达到SOTA,领先Claude Code和Codex,覆盖美国年收入2.
4万亿美元的岗位需求。D1自动决定每个步骤由哪个智能体执行,替代手动比较多个agent输出。
This might be big Instead of just sending one prompt to multiple agents and comparing the results manually... this D1 layer is: - Deciding which agent acts at each step - Only 3B active parameters - SOTA on router benchmarks So it just acts between you and the different agents Pretty cool!
This report is categorized under AITOP's "Models" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Offloop, D1, GDPval, Claude Code.
Source attribution: curated from Paul Couvert (@itsPaulAi). Cross-source corroboration: 4 independent sources have covered this development (including elvis, 宝玉, arXiv: DeepSeek), confirming industry-wide signal strength.