# HyperEyes: Dual-Grained Efficiency-Aware Reinforcement Learning for Parallel Multimodal Search Agents

arXiv: https://arxiv.org/abs/2605.07177 · HTML: https://mlsanity.com/paper/2605.07177/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge- and tool-augmented reasoning, one of 17 directions in LLM reasoning.
- Builds on 34 of its 44 references that are in the map.
- First posted on arXiv 8 May 2026.

## Placed in

- LLM reasoning / [Knowledge- and tool-augmented reasoning](https://mlsanity.com/lr/knowledge-and-tool-augmented-reasoning/)

## Builds on (most cited first, 30 of 34)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://mlsanity.com/paper/1909.08053/) (2019)
- [Fine-Grained Visual Classification of Aircraft](https://mlsanity.com/paper/1306.5151/) (2013)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [PaLM 2 Technical Report](https://mlsanity.com/paper/2305.10403/) (2023)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [Measuring and Narrowing the Compositionality Gap in Language Models](https://mlsanity.com/paper/2210.03350/) (2022)
- [Kimi K2.5: Visual Agentic Intelligence](https://mlsanity.com/paper/2602.02276/) (2026)
- [SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning](https://mlsanity.com/paper/2408.05517/) (2024)
- [Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval](https://mlsanity.com/paper/2004.01804/) (2020)
- [A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?](https://mlsanity.com/paper/2503.24235/) (2025)
- [DeepEyesV2: Toward Agentic Multimodal Model](https://mlsanity.com/paper/2511.05271/) (2025)
- [MMSearch-R1: Incentivizing LMMs to Search](https://mlsanity.com/paper/2506.20670/) (2025)
- [WebWatcher: Breaking New Frontier of Vision-Language Deep Research Agent](https://mlsanity.com/paper/2508.05748/) (2025)
- [MMSearch: Benchmarking the Potential of Large Models as Multi-modal Search Engines](https://mlsanity.com/paper/2409.12959/) (2024)
- [MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling](https://mlsanity.com/paper/2511.11793/) (2025)
- [DeepDive: Advancing Deep Search Agents with Knowledge Graphs and Multi-Turn RL](https://mlsanity.com/paper/2509.10446/) (2025)
- [REDSearcher: A Scalable and Cost-Efficient Framework for Long-Horizon Search Agents](https://mlsanity.com/paper/2602.14234/) (2026)
- [DeepMMSearch-R1: Empowering Multimodal LLMs in Multimodal Web Search](https://mlsanity.com/paper/2510.12801/) (2025)
- [MMSearch-Plus: Benchmarking Provenance-Aware Search for Multimodal Browsing Agents](https://mlsanity.com/paper/2508.21475/) (2025)
- [Vision-DeepResearch: Incentivizing DeepResearch Capability in Multimodal Large Language Models](https://mlsanity.com/paper/2601.22060/) (2026)
- [INQUIRE: A Natural World Text-to-Image Retrieval Benchmark](https://mlsanity.com/paper/2411.02537/) (2024)
- [Skywork-R1V4: Toward Agentic Multimodal Intelligence through Interleaved Thinking with Images and DeepResearch](https://mlsanity.com/paper/2512.02395/) (2025)
- [ParallelSearch: Train your LLMs to Decompose Query and Search Sub-queries in Parallel with Reinforcement Learning](https://mlsanity.com/paper/2508.09303/) (2025)

## Built on by (most cited first, 10 of 10)

- [VistaHop: Benchmarking Long-Horizon Visual DeepSearch](https://mlsanity.com/paper/2606.03273/) (2026)
- [StreamArena: Toward Continuous, Interactive, and Long-Horizon Agentic Streaming Video Understanding](https://mlsanity.com/paper/2608.05703/) (2026)
- [SimpleSearch-VL: A Simple Recipe for Multimodal Agentic Deep Search](https://mlsanity.com/paper/2606.31504/) (2026)
- [WeAgent-MMSearch: Native Text-Vision Interaction for Multimodal Search Agents](https://mlsanity.com/paper/2608.28062/) (2026)
- [HindSearch: Trajectory-Level Hindsight Critique for Search-Augmented Reinforcement Learning](https://mlsanity.com/paper/2608.01597/) (2026)
- [Thinking Beyond Videos: Unifying Video Reasoning and Deep Research for Open-World Video Agents](https://mlsanity.com/paper/2608.23329/) (2026)
- [AdaVDR: Adaptive Tool Use and Reflection for Video Deep Research](https://mlsanity.com/paper/2608.25559/) (2026)
- [Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents](https://mlsanity.com/paper/2610.01892/) (2026)
- [SCAD: Structured Credit Assignment and Distillation for Long-Horizon Agents](https://mlsanity.com/paper/2610.03372/) (2026)
- [Long-MDR: Long-Context Reinforcement Learning for Multimodal Deep-Research Agents](https://mlsanity.com/paper/2610.05195/) (2026)

## Abstract (arXiv)

Existing multimodal search agents process target entities sequentially, issuing one tool call per entity and accumulating redundant interaction rounds whenever a query decomposes into independent sub-retrievals. We argue that effective multimodal agents should search wider rather than longer: dispatching multiple grounded queries concurrently within a round. To this end, we present HyperEyes, a parallel multimodal search agent that fuses visual grounding and retrieval into a single atomic action, enabling concurrent search across multiple entities while treating inference efficiency as a first-class training objective. HyperEyes is trained in two stages. For cold-start supervision, we develop a Parallel-Amenable Data Synthesis Pipeline covering visual multi-entity and textual multi-constraint queries, curating efficiency-oriented trajectories via Progressive Rejection Sampling. Building on this, our central contribution, a Dual-Grained Efficiency-Aware Reinforcement Learning framework, operates at two levels. At the macro level, we propose TRACE (Tool-use Reference-Adaptive Cost Efficiency), a trajectory-level reward whose reference is monotonically tightened during training to suppress superfluous tool calls without restricting genuine multi-hop search. At the micro level, we adapt On-Policy Distillation to inject dense token-level corrective signals from an external teacher on failed rollouts, mitigating the credit-assignment deficiency of sparse outcome rewards. Since existing benchmarks evaluate accuracy as the sole metric, omitting inference cost, we introduce IMEB, a human-curated benchmark of 300 instances that jointly evaluates search capability and efficiency. Across six benchmarks, HyperEyes-30B surpasses the strongest comparable open-source agent by 9.9% in accuracy with 5.3x fewer tool-call rounds on average.