Decentralized AI infrastructure firm Gata has launched the ChatGPT-RealUser-2.2M dataset, featuring over 2.24 million real conversations and nearly 3.56 million Q&A pairs. Collected through Gata's GPT-to-Earn program, the dataset includes interactions with GPT-3.5, GPT-4, and o1 from 2024 to 2025, involving more than 15,000 users. Notably, it is twice the size of previous datasets by the Allen Institute for AI and includes substantial crypto-related content due to its on-chain incentive mechanism. A preview with 600 conversation samples is available on Hugging Face, while the full dataset is intended for research and commercial applications. This release follows Gata's $4 million seed funding round in May 2025, backed by YZi Labs and IDG Blockchain.