Towards On-Policy Data Evolution for Visual-Native Multimodal Deep Search Agents

TL;DR

This paper introduces a visual-native agent with an image bank protocol and an on-policy data evolution framework, significantly improving multimodal deep search performance across multiple benchmarks.

Contribution

It proposes a novel image bank reference protocol and a closed-loop data refinement method that enhances training data relevance and reusability for multimodal agents.

Findings

ODE improves Qwen3-VL-8B from 24.9% to 39.0% accuracy.

ODE surpasses Gemini-2.5 Pro in standard setting (37.9%).

Image-bank reuse benefits complex iterative visual tasks.

Abstract

Multimodal deep search requires an agent to solve open-world problems by chaining search, tool use, and visual reasoning over evolving textual and visual context. Two bottlenecks limit current systems. First, existing tool-use harnesses treat images returned by search, browsing, or transformation as transient outputs, so intermediate visual evidence cannot be re-consumed by later tools. Second, training data is usually built by fixed curation recipes that cannot track the target agent's evolving capability. To address these challenges, we first introduce a visual-native agent harness centered on an image bank reference protocol, which registers every tool-returned image as an addressable reference and makes intermediate visual evidence reusable by later tools. On top of this harness, On-policy Data Evolution (ODE) runs a closed-loop data generator that refines itself across rounds from rollouts of the policy being trained. This per-round refinement makes each round's data target what the current policy still needs to learn. The same framework supports both diverse supervised fine-tuning data and policy-aware reinforcement learning data curation, covering the full training lifecycle of the target agent. Across 8 multimodal deep search benchmarks, ODE improves the Qwen3-VL-8B agent from 24.9% to 39.0% on average, surpassing Gemini-2.5 Pro in standard agent-workflow setting (37.9%). At 30B, ODE raises the average score from 30.6% to 41.5%. Further analyses validate the effectiveness of image-bank reuse, especially on complex tasks requiring iterative visual refinement, while rollout-feedback evolution yields more grounded SFT traces and better policy-matched RL tasks than static synthesis.