Dataset guide
Match the collection and filtering to the research question.
Start with the open WildChat-4.8M release. Full releases include flagged content and require an access request. For replication, match the repository, revision, and filters used by the original study.
Current downloads and access requests: yuntian-deng/WildChat-4.8M and yuntian-deng/WildChat-4.8M-Full. Request Full access in the latter repository.
Counts and access information checked October 19, 2026 against the linked dataset cards.
Open download; toxicity-filtered
The expanded human–ChatGPT collection, excluding conversations flagged by OpenAI Moderation or Detoxify.
Gated; access request required
Includes flagged and unflagged conversations, excluding the sexual/minors category. Contains 122,921 o1-mini/o1-preview conversations.
Use the yuntian-deng repositories linked here for downloads and access requests. To reproduce an earlier study, match its collection window, revision, and filters. Source.
Submit an access request on yuntian-deng/WildChat-4.8M-Full explaining why your work needs flagged content. Requests are reviewed manually. Source.
The processing pipeline removes conversations reserved for WildBench. Preserve that separation when constructing training sets or comparing with the benchmark. Source.
WildChat-50M, by Benjamin Feuer and Chinmay Hegde, adds responses from more than 50 open-weight models to WildChat prompts. Source.
Cite Zhao, Ren, Hessel, Cardie, Choi, and Deng’s ICLR 2024 WildChat paper for the dataset. Cite the WildVis paper when using the visualizer, and identify the downloaded release and revision. Source.
The Full release card lists ODC-BY and explains how to request removal using conversation hashes or turn identifiers and the contact details on Yuntian Deng’s homepage. Source.
Geographic locations are inferred from IP addresses. Section 6 of the paper describes the service’s user population and sampling biases. Source.
In the expanded releases, conversation_hash is a content hash and can repeat across conversations. turn_identifier uniquely identifies each turn. Source.
Read the ODC-BY license and the Full release card for terms and collection details. For removal requests, follow the dataset card’s instructions.
The original paper describes an approximately one-million-conversation corpus. The 4.8M releases extend the collection through July 2025 and add more models.
Start of the collection window reported for the expanded corpus in the May 2024 paper. Source.
Collection start.
The WildChat arXiv version describes the expanded one-million-conversation corpus. Source.
First arXiv version.
The 4.8M release update extends coverage to before August 1, 2025 and adds secret scanning. Source.
Collection through July 31, 2025.