---

About Tuchong AI · 关于图虫 AI

Tuchong (图虫) is a leading visual content platform in China. Building on years of expertise in visual content management, copyright-compliant processing, and visual data understanding, Tuchong AI is committed to constructing the data infrastructure for large-scale AI models — providing high-quality, diverse, and trustworthy data resources for the next generation of AI.

图虫（Tuchong） 是中国领先的视觉内容平台，长期深耕高质量视觉内容生态。依托在图片与视频内容管理、版权合规处理、视觉数据理解等领域的多年积累，图虫 AI 致力于建设大模型时代的数据基础设施，为下一代 AI 模型提供高质量、多样化、可信赖的数据资源。

---

Focus Areas · 重点方向

01 · Multimodal Datasets · 多模态数据集

Curated, high-quality datasets for training and evaluating multimodal models:

面向多模态模型训练与评测的高质量数据集：

• Vision-Language Models (VLM) · 视觉语言模型
• Image Understanding & Generation · 图像理解与生成
• Multimodal Reasoning · 多模态推理
• AI Agent Training · 智能体训练数据

02 · Agent Trajectory Data · 智能体轨迹数据

As AI agents evolve, models must learn not only knowledge, but how to complete complex, long-horizon tasks. We build real, high-quality agent interaction trajectories covering:

随着智能体的发展，模型不仅需要学习知识，更需要学会完成复杂任务。我们构建真实、高质量的智能体交互轨迹数据：

• Long-horizon task execution · 长程任务执行轨迹
• Agent interaction records · 智能体操作过程记录
• Tool invocation patterns · 工具调用行为
• Environment feedback · 环境反馈信息
• Planning & execution processes · 任务规划与执行过程

03 · Data Quality & Evaluation · 数据质量与评测

We maintain rigorous quality control across the full data lifecycle, grounded in systematic evaluation:

我们围绕数据全生命周期建立严格的质量控制与评测体系：

• Data quality benchmarks · 数据质量基准
• Automated + human-in-the-loop evaluation · 自动化评测与人工审核相结合
• Benchmark construction · Benchmark 构建

---

Data Infrastructure · 数据生产基础设施

Every dataset is backed by an end-to-end production pipeline:

每份数据集均由端到端的数据生产流程保障：

High-quality data infrastructure is the foundation of reliable AI.

高质量的数据基础设施，是可靠人工智能发展的基石。

---

Open Source · 开源计划

Tuchong AI is committed to contributing to the global AI community through open collaboration. Datasets are released as free demo data, with full datasets available under paid licensing for both academic and commercial use — see Data Access.

图虫 AI 致力于与全球研究者、开发者共建社区、共同推动人工智能的发展。数据集以「样例免费、全量付费」的方式发布：样例数据供学术研究免费使用，完整数据集学术与商业用途均需付费授权，详见 Data Access · 数据获取。

Featured releases · 已发布数据集：

More datasets, benchmarks, and data-processing tools are on the way.

更多数据集、Benchmark 与数据处理工具持续发布中。

---

Data Access · 数据获取

Dataset cards are publicly visible. The demo subset is free for academic research and evaluation; the full dataset is available under paid licensing — for both academic and commercial use.

Dataset Card 公开可见；样例数据（Demo）免费供学术研究与评估使用；完整数据集需付费授权获取，学术与商业用途均需购买。

Request Workflow · 数据申请流程

1. View — read the README / Dataset Card and try the demo data · 阅读 README / Dataset Card，试用样例数据
2. Apply — send an email stating your institution, purpose, and target dataset · 发送申请邮件，注明机构、用途与目标数据集
3. Review — the Tuchong AI data team evaluates and replies with licensing terms · 图虫 AI 数据团队审核，并回复授权条款与报价
4. Sign — sign the applicable agreement (research or commercial license) · 签署对应协议（学术研究协议 / 商业授权协议）
5. Access — access is granted upon license and payment completion · 完成签约与付款后，通过邮件获得数据访问权限

Application email must include · 申请邮件须包含：

• Name & institution · 姓名与所属机构
• Role / title · 职位
• Purpose: academic research or commercial use · 用途：学术研究或商业用途
• Target dataset · 目标数据集

License highlights · 授权核心条款：

• Demo data is limited to academic research and evaluation · 样例数据仅限学术研究与评估使用
• No redistribution or re-release · 未经授权不得再分发或再发布
• Citation of Tuchong AI datasets is required · 须规范引用图虫 AI 数据集

---

Collaboration · 合作交流

We welcome collaboration with AI researchers, open-source communities, universities, research institutions, and AI model & data teams — for dataset cooperation, research collaboration, technical exchange, and commercial partnerships.

我们欢迎 AI 研究人员、开源社区、高校及科研机构、AI 模型开发与数据研究团队与我们联系，探讨数据集合作、AI 研究合作、技术交流与商业合作。

Contact · 联系方式

---