paperarXivTrust 82 · PrimaryPublished 4d agoLive · 3d ago
Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent
We introduce Agents-A1, a 35B Mixture-of-Experts Agentic Model that reaches trillion-parameter-level performance by scaling the agent horizon. We investigate agent-horizon scaling from two perspectives: scaling long-horizon trajectories and scaling heterogeneous agent abilities. To support this goal, we build a long-horizon knowledge-action infrastructure that connects external knowledge, actions, observations, and verifier outcomes, producing agentic trajectories with an average length of 45K tokens. Based on this, we train Agents-A1 with a three-stage recipe. First, we perform full-domain su
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Has model
Covers
newsGeneral Intuition’s $2.3B bet that video games can train AI agents for the real worldnewsAlibaba's model never trained as an agent — and improved agent performance across seven benchmarksnewsGoogle DeepMind is worried about what happens when millions of agents start to interactnewsMonitor and debug generative AI inference with SageMaker detailed metrics and Insights dashboard on CloudWatch
Implements
Covers (incoming)
Implements (incoming)
Related across the graph
newsInto the Omniverse: Three Workflows for Improving Vision AI Agent Accuracy With Synthetic Data and Fine-TuningnewsDeepSeek open sources DSpark, a new framework to speed up LLM inference by up to 85%newsGeneral Intuition’s $2.3B bet that video games can train AI agents for the real worldrepotensorflow/servingnewsClaude Meets Blackwell Ultra: Anthropic’s Models Now Run on NVIDIA GB300 in AzurerepoFabiojvv/ai-cortex-hubnewsMonitor and debug generative AI inference with SageMaker detailed metrics and Insights dashboard on CloudWatchnewsGoogle DeepMind is worried about what happens when millions of agents start to interactrepodeepset-ai/haystack-core-integrationsnewsAlibaba's model never trained as an agent — and improved agent performance across seven benchmarksmodelAgentCore-8Breporeunios2024/cortex-sentinel-trading-nexus