DeepSearch-World: Self-Distillation for Deep Search Agents in a Verifiable Environment
Summary
DeepSearch-Evolve introduces a self-distillation framework for web agents using a verifiable environment (DeepSearch-World) with 420K multi-hop QA tasks, achieving competitive performance without distillation from stronger models.
View Cached Full Text
Cached at: 07/10/26, 06:11 AM
# DeepSearch-World: Self-Distillation for Deep Search Agents in a Verifiable Environment Source: [https://arxiv.org/abs/2607.07820](https://arxiv.org/abs/2607.07820) [View PDF](https://arxiv.org/pdf/2607.07820) > Abstract:Training tool\-use agents to improve from their own experience remains challenging, as supervised fine\-tuning relies on fixed teacher\-distilled trajectories, while sparse\-reward reinforcement learning provides weak supervision for long\-horizon interactions\. We present DeepSearch\-Evolve, a self\-distillation framework for web agents built on DeepSearch\-World, a deterministic and verifiable environment with reproducible search and page\-reading tools\. DeepSearch\-World contains 420K multi\-hop QA tasks constructed from entity\-level random walks and supports key agentic cognitive behaviors useful for self\-evolving, including progress verification, grounded reflection, and failure recovery\. DeepSearch\-Evolve iteratively performs trajectory generation, filtering, data mixing, and fine\-tuning to train stronger agents\. Without distillation from more capable models, DeepSearch\-World\-9B achieves competitive performance compared with open\-source agents, reaching 31\.2% on BrowseComp, 61\.5% on GAIA, and 93\.4% on HotpotQA, showing that verifiable environments enable scalable self\-evolution for long\-horizon web agents\. We will release the environment, 420K training pool, validation set, model, and code to facilitate future research on self\-improving deep search agents\. ## Submission history From: Xinyu Geng \[[view email](https://arxiv.org/show-email/31d1c647/2607.07820)\] **\[v1\]**Wed, 8 Jul 2026 18:03:41 UTC \(2,480 KB\)
Similar Articles
@DanKornas: DeepDive is a pattern for deep search agents: synthesize QA from knowledge graphs, then train multi-turn browsing with …
DeepDive is a pattern for building deep search agents that synthesizes QA from knowledge graphs and trains multi-turn browsing with reinforcement learning (GRPO). It includes entity obfuscation and test-time scaling with tool calls.
Training LLM Agents for Spontaneous, Reward-Free Self-Evolution via World Knowledge Exploration
This paper proposes a method to train LLM agents with intrinsic meta-evolution capabilities, enabling spontaneous self-improvement without external rewards at inference time. Applied to Qwen3-30B and Seed-OSS-36B, the approach yields a 20% performance boost on web navigation benchmarks, with a 14B model outperforming Gemini-2.5-Flash.
Echoverse: Deep, Evolving Environments for Training Computer-Use Agents at Scale
Echoverse presents a method for generating deep, evolving synthetic environments to train computer-use agents, demonstrating substantial accuracy gains and releasing a benchmark with grounded graders.
Self-Verified Distillation: Your Language Model Is Secretly Its Own Synthetic Data Pipeline
Proposes Self-Verified Distillation, a method where LLMs generate and self-verify candidate solutions from unlabeled seed questions using prompt-based verification, then train on the filtered dataset, achieving significant gains on math, science, and coding benchmarks across Qwen3 models.
Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
Agent-World introduces a self-evolving training framework for general agent intelligence that autonomously discovers real-world environments and tasks via the Model Context Protocol, enabling continuous learning. Agent-World-8B and 14B models outperform strong proprietary models across 23 challenging agent benchmarks.