Sage-A1.5: From Long-Horizon Execution to Scientific Exploration with Capability Composition
What this work does
Sage-A1.5 develops a 35B mixture-of-experts agent whose learned orchestration and verification behaviors support sustained scientific investigation across tools and parallel research tasks.
THE RESEARCH CONVERSATION
Discussion & reactions
Share a question, an insight, or a replication experience. Sign in with GitHub to join the conversation.