Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent

• URL: http://arxiv.org/abs/2606.30616v1
• Date: Mon, 29 Jun 2026 17:50:54 GMT
• Title: Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent
• Abstract summary: We introduce Agents-A1, a 35B Mixture-of-Experts Agentic Model that reaches trillion- parameter-level performance by scaling the agent horizon.<n>We build a long-horizon knowledge-action infrastructure that connects external knowledge, actions, observations, and verifier outcomes.<n>Agents-A1 achieves strong and broad performance for long-horizon agent benchmarks.
• Score: 98.33217711280383
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: 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 supervised fine-tuning to align the base model with broad agentic behaviors. Second, we train domain-level teacher models to capture specialized expertise in each domain. Third, we propose a multi-teacher domain-routed on-policy distillation with salient vocabulary alignment to improve knowledge transfer efficiency across different domains, unifying six heterogeneous domains into one deployable student model. Agents-A1 achieves strong and broad performance for long-horizon agent benchmarks. Compared with 1T-parameter model such as Kimi-K2.6 and DeepSeek-V4-pro, Agents-A1 achieves leading results on SEAL-0 (56.4), IFBench (80.6), HiPhO (46.4), FrontierScience-Olympiad (79.0), and MolBench-Bind (56.8), and remains highly competitive on SciCode (44.3), HLE (47.6) and BrowseComp (75.5). We hope this work provides the community with a practical path for scaling the horizon using a 35B agent that can reach or match the performance of 1T models on long-horizon tasks.

Related papers

• AgentOmnia: Scaling Agentic Models for Full-Scenario Applications [36.656762500581216]
  Large language model agents have advanced rapidly, yet progress remains fragmented across domains, capabilities, task difficulty, and interaction settings.<n>We frame this as full-scenario agentic scaling and present Agent Omnia, a framework coordinating task-space definition, data synthesis, post-training, evaluation, and improvement across To-Consumer (ToC), To-Business (ToB), and To-Employee (ToE) applications.
  arXiv  Detail & Related papers  (2026-07-25T09:58:24Z)
• Structured Distillation of Web Agent Capabilities Enables Generalization [33.41723014134473]
  We introduce Agent-as-Annotators, a framework that structures synthetic trajectory generation for web agents by analogy to human annotation roles.<n>Using Gemini 3 Pro as teacher, we generate 3,000 trajectories across six web environments.<n>The resulting model achieves 41.5% on WebArena, surpassing closed-source models such as Claude 3.5 Sonnet (36.0%) and GPT-4o (31.5%) under the same evaluation protocol.
  arXiv  Detail & Related papers  (2026-04-09T04:04:15Z)
• Yet Even Less Is Even Better For Agentic, Reasoning, and Coding LLMs [28.672417618778976]
  Training effective software engineering agents requires large volumes of task-specific trajectories.<n>We propose an end-to-end training framework that achieves superior agentic capabilities with fewer but higher-quality training trajectories.
  arXiv  Detail & Related papers  (2026-04-01T12:33:25Z)
• SciAgentGym: Benchmarking Multi-Step Scientific Tool-use in LLM Agents [100.12367115920121]
  We introduce SciGymAgent, a scalable interactive environment featuring 1,780 domain-specific tools across four natural science disciplines.<n>We also present SciAgentBench, a tiered evaluation suite designed to stress-test agentic capabilities.
  arXiv  Detail & Related papers  (2026-02-13T14:58:18Z)
• GenAgent: Scaling Text-to-Image Generation via Agentic Multimodal Reasoning [54.42973725693]
  We introduce GenAgent, unifying visual understanding and generation through an agentic multimodal model.<n>GenAgent significantly boosts base generator(FLUX.1-dev) performance on GenEval++ and WISE.<n>Our framework demonstrates three key properties: 1) cross-tool generalization to generators with varying capabilities, 2) test-time scaling with consistent improvements across interaction rounds, and 3) task-adaptive reasoning that automatically adjusts to different tasks.
  arXiv  Detail & Related papers  (2026-01-26T14:49:04Z)
• APIGen-MT: Agentic Pipeline for Multi-Turn Data Generation via Simulated Agent-Human Interplay [86.01901238059261]
  APIGen-MT is a framework that generates verifiable and diverse multi-turn agent data.<n>We train a family of models -- the xLAM-2-fc-r series with sizes ranging from 1B to 70B parameters.<n>Our models outperform frontier models such as GPT-4o and Claude 3.5 on $tau$-bench and BFCL benchmarks.
  arXiv  Detail & Related papers  (2025-04-04T17:13:57Z)
• AgentBank: Towards Generalized LLM Agents via Fine-Tuning on 50000+ Interaction Trajectories [44.68479398604844]
  Fine-tuning on agent-environment interaction trajectory data holds significant promise for surfacing generalized agent capabilities. We introduce AgentBank, by far the largest trajectory tuning data collection featuring more than 50k diverse high-quality interaction trajectories.
  arXiv  Detail & Related papers  (2024-10-10T08:19:12Z)
• xLAM: A Family of Large Action Models to Empower AI Agent Systems [111.5719694445345]
  We release xLAM, a series of large action models designed for AI agent tasks. xLAM consistently delivers exceptional performance across multiple agent ability benchmarks.
  arXiv  Detail & Related papers  (2024-09-05T03:22:22Z)
• Lory: Fully Differentiable Mixture-of-Experts for Autoregressive Language Model Pre-training [73.90260246781435]
  We present Lory, the first approach that scales such architectures to autoregressive language model pre-training. We show significant performance gains over parameter-matched dense models on both perplexity and a variety of downstream tasks. Despite segment-level routing, Lory models achieve competitive performance compared to state-of-the-art MoE models with token-level routing.
  arXiv  Detail & Related papers  (2024-05-06T03:06:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.