# SciAgentGym: Benchmarking Multi-Step Scientific Tool-use in LLM Agents

Source: [SciAgentGym: Benchmarking Multi-Step Scientific Tool-use in LLM Agents](<https://www.sota2.com/research/paper/6992a47b9ebc499c97406c24>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-02T05:00:51.557Z

Authors: Yujiong Shen, Yajie Yang, Zhiheng Xi, Binze Hu, Huayu Sha, Jiazheng Zhang, Qiyuan Peng, Junlin Shang, Jixuan Huang, Yutao Fan, Jingqi Tong, Shihan Dou, Ming Zhang, Lei Bai, Zhenfei Yin, Tao Gui, Xingjun Ma, Qi Zhang, Xuanjing Huang, Yu-Gang Jiang

Published: 2026-02-13T14:58:18.000Z

## Abstract

Scientific reasoning inherently demands integrating sophisticated toolkits to navigate domain-specific knowledge. Yet, current benchmarks largely overlook agents' ability to orchestrate tools for such rigorous workflows. To bridge this gap, we introduce SciAgentGym, a scalable interactive environment featuring 1,780 domain-specific tools across four natural science disciplines, supported by a robust execution infrastructure. Complementing this, we present SciAgentBench, a tiered evaluation suite designed to stress-test agentic capabilities from elementary actions to long-horizon workflows. Our evaluation identifies a critical bottleneck: state-of-the-art models still struggle with complex scientific tool-use, and their performance degrades substantially as interaction horizons extend. To address this, we propose SciForge, a data synthesis method that models the tool action space as a dependency graph to generate logic-aware training trajectories. By fine-tuning on these trajectories, our SciAgent-8B outperforms the significantly larger Qwen3-VL-235B-Instruct while exhibiting positive cross-domain transfer of scientific tool-use capabilities. These results underscore the promising potential of next-generation autonomous scientific agents.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Multi-step Scientific Tool-use on SciAgentBench Main Results 1.0](<https://www.sota2.com/research/sota/multi-step-scientific-tool-use-on-sciagentbench-main-results-1-0.md>): Multi-step Scientific Tool-use; SciAgentBench Main Results 1.0; SR (w/o Tools): 23.3.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2602.12984>)

[PDF](<https://arxiv.org/pdf/2602.12984.pdf>)

[Implementation](<https://github.com/CMarsRover/SciAgentGYM>)