# AgentCPM-Explore: Realizing Long-Horizon Deep Exploration for Edge-Scale Agents

arXiv: https://arxiv.org/abs/2602.06485 · HTML: https://mlsanity.com/paper/2602.06485/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use.
- Builds on 12 of its 21 references that are in the map.
- First posted on arXiv 6 Feb 2026.

## Placed in

- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 12 of 12)

- [DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models](https://mlsanity.com/paper/2512.02556/) (2025)
- [GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models](https://mlsanity.com/paper/2508.06471/) (2025)
- [WebSailor: Navigating Super-human Reasoning for Web Agent](https://mlsanity.com/paper/2507.02592/) (2025)
- [WebDancer: Towards Autonomous Information Seeking Agency](https://mlsanity.com/paper/2505.22648/) (2025)
- [LLM in a flash: Efficient Large Language Model Inference with Limited Memory](https://mlsanity.com/paper/2312.11514/) (2023)
- [Beyond Ten Turns: Unlocking Long-Horizon Agentic Search with Large-Scale Asynchronous RL](https://mlsanity.com/paper/2508.07976/) (2025)
- [DELLA-Merging: Reducing Interference in Model Merging through Magnitude-Based Sampling](https://mlsanity.com/paper/2406.11617/) (2024)
- [MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling](https://mlsanity.com/paper/2511.11793/) (2025)
- [WebLeaper: Empowering Efficiency and Efficacy in WebAgent via Enabling Info-Rich Seeking](https://mlsanity.com/paper/2510.24697/) (2025)
- [CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge](https://mlsanity.com/paper/2506.02847/) (2025)
- [Pushing up to the Limit of Memory Bandwidth and Capacity Utilization for Efficient LLM Decoding on Embedded FPGA](https://mlsanity.com/paper/2502.10659/) (2025)
- [Nested Browser-Use Learning for Agentic Information Seeking](https://mlsanity.com/paper/2512.23647/) (2025)

## Built on by (most cited first, 12 of 12)

- [AgentProcessBench: Diagnosing Step-Level Process Quality in Tool-Using Agents](https://mlsanity.com/paper/2603.14465/) (2026)
- [LiteResearcher: A Scalable Agentic RL Training Framework for Deep Research Agent](https://mlsanity.com/paper/2604.17931/) (2026)
- [DR-Venus: Towards Frontier Edge-Scale Deep Research Agents with Only 10K Open Data](https://mlsanity.com/paper/2604.19859/) (2026)
- [GUI Agents with Reinforcement Learning: Toward Digital Inhabitants](https://mlsanity.com/paper/2604.27955/) (2026)
- [ABSeeker: Training Long-Horizon Search Agents via Answer-Backtracked Credit Assignment](https://mlsanity.com/paper/2608.05102/) (2026)
- [When Deep Research Agents Stagnate: Enhancing Reasoning with Retrieval-Aware Agent Control](https://mlsanity.com/paper/2608.15191/) (2026)
- [Beyond Reward Engineering: A Data Recipe for Long-Context Reinforcement Learning](https://mlsanity.com/paper/2606.18831/) (2026)
- [Inducing Process Supervision from Outcome-Only Reinforcement Learning](https://mlsanity.com/paper/2609.36641/) (2026)
- [TIDE-Bench: Task-Aware and Diagnostic Evaluation of Tool-Integrated Reasoning](https://mlsanity.com/paper/2605.09544/) (2026)
- [Boosting Deepresearch and LongContext Ability with Self-Generated Deepresearch Rollouts Traces](https://mlsanity.com/paper/2609.20844/) (2026)
- [IterSynth: Rethinking Deep Search Agents via Role-Decoupled Iterative Synthesis](https://mlsanity.com/paper/2609.29444/) (2026)
- [VACE: Validation-Gated Alternating Co-Evolution of Agent Models and Harnesses](https://mlsanity.com/paper/2609.37105/) (2026)

## Abstract (arXiv)

While Large Language Model (LLM)-based agents have shown remarkable potential for solving complex tasks, existing systems remain heavily reliant on large-scale models, leaving the capabilities of edge-scale models largely underexplored. In this paper, we present the first systematic study on training agentic models at the 4B-parameter scale. We identify three primary bottlenecks hindering the performance of edge-scale models: catastrophic forgetting during Supervised Fine-Tuning (SFT), sensitivity to reward signal noise during Reinforcement Learning (RL), and reasoning degradation caused by redundant information in long-context scenarios. To address the issues, we propose AgentCPM-Explore, a compact 4B agent model with high knowledge density and strong exploration capability. We introduce a holistic training framework featuring parameter-space model fusion, reward signal denoising, and contextual information refinement. Through deep exploration, AgentCPM-Explore achieves state-of-the-art (SOTA) performance among 4B-class models, matches or surpasses 8B-class SOTA models on four benchmarks, and even outperforms larger-scale models such as Claude-4.5-Sonnet or DeepSeek-v3.2 in five benchmarks. Notably, AgentCPM-Explore achieves 97.09% accuracy on GAIA text-based tasks under pass@64. These results provide compelling evidence that the bottleneck for edge-scale models is not their inherent capability ceiling, but rather their inference stability. Based on our well-established training framework, AgentCPM-Explore effectively unlocks the significant, yet previously underestimated, potential of edge-scale models.