danish-foundation-models/ai-arenaen-conversations

AI Arenaen Conversations

A large dataset of conversations from AI-Arenaen, the Danish subset of the compar:IA platform.

Origin of the data: what is AI-Arenaen?

The conversations are collected using AI-Arenaen, the Danish entry point to the compar:IA platform, which is a Conversational AI comparison tool (a "chatbot arena"), developed within the French Ministry of Culture and adapted for Danish users by Danish Foundation Models and The ministry of digital affair. Its dual mission is:

• To educate and raise awareness about the diversity of models, cultural and linguistic biases, and the environmental impact of conversational AIs.
• To improve conversational AIs by publishing alignment datasets and by creating a ranking of conversational AI models (in progress).

The platform is built on compar:IA (incubated by Atelier numérique and AllIAnce), integrated into the beta.gouv.fr program of the Interministerial Digital Directorate (DINUM). The source code for AI-Arenaen is publicly available on GitHub.

Dataset content

The dataset includes results generated by more than 100 conversational models, both open-source and proprietary, of various sizes. Questions are asked by Danish users and correspond to real, unconstrained uses. The dataset is continuously updated as new conversations are collected.

Columns of the ai-arenaen-conversations dataset

Column	Description
id	Unique identifier for each entry in the dataset
timestamp	Conversation timestamp
model_a_name	Name of the first model
model_b_name	Name of the second model
conversation_a	Full structure of the conversation with the first model
conversation_b	Full structure of the conversation with the second model
conv_turns	Number of dialogue turns in the conversation
conversation_pair_id	Unique identifier for the pair of conversations
conv_a_id	Unique identifier for the conversation with the first model
conv_b_id	Unique identifier for the conversation with the second model
session_hash	User session identifier
visitor_id	Unique anonymized identifier for the user
country	Country – empty for now
city	City – empty for now
model_pair_name	Set representation of the two compared models
opening_msg	First message sent by the user
system_prompt_a	System prompt for model A
system_prompt_b	System prompt for model B
selected_category	Prompt suggestion category chosen by the user (if a suggested prompt was used)
is_unedited_prompt	Indicates whether the suggested prompt was used as-is
mode	Indicates the selected mode
custom_models_selection	If the user manually chose models, which ones were selected
short_summary	Brief summary of the conversation content
keywords	List of keywords summarizing the content
categories	Thematic categories of each conversation
languages	List of detected languages in the conversation
total_conv_a_output_tokens	Total number of output tokens generated by model A during the conversation
total_conv_b_output_tokens	Total number of output tokens generated by model B during the conversation
model_a_total_params	Estimated or known total size of model A
model_a_acitve_params	Estimated or known number of active parameters in model A
model_b_total_params	Estimated or known total size of model B
model_b_acitve_params	Estimated or known number of active parameters in model B
total_conv_a_kwh	Estimated electricity consumption in kWh for model A
total_conv_b_kwh	Estimated electricity consumption in kWh for model B

Purpose of this dataset

We make this dataset available to model developers as well as the research community in artificial intelligence and social sciences to support progress in the following areas:

• Training and alignment of conversational language models, particularly for Danish
• Human–machine interactions and the specific behaviors involved in conversational AI systems
• Improvement of LLM evaluation methods
• AI safety and content moderation

If you use the AI-Arenaen dataset, we would love to hear about your use cases and feedback. It will help us improve the re-use experience.

Data post-processing

User consent is collected through the "Terms of use" section on the website. A detection of personally identifiable information (PII) was carried out (results are shown in the contains_pii column of the dataset), and conversations containing such information were anonymized. However, we do not apply any filtering or moderation of potentially toxic or hateful content, to allow researchers to study safety issues related to LLM use in real-world contexts.

Licenses

The dataset is released under the Etalab 2.0 license, as well as the compatible Creative Commons Attribution 4.0 (CC BY 4.0).

Citation

If you use this dataset, please cite:

@misc{aiarenaen-conversations,
  title        = {AI-Arenaen: A Danish Chatbot Arena Dataset},
  year         = {2026},
}

Related datasets

Dataset	Description	Users	Link
ai-arenaen-conversations	Conversation pairs between two models (this dataset)	🇩🇰 Danish	View
ai-arenaen-reactions	User reactions at the message level	🇩🇰 Danish	View
ai-arenaen-votes	User preferences after full conversations	🇩🇰 Danish	View
comparia-conversations	Conversation pairs between two models	🇫🇷 French	View
comparia-reactions	User reactions at the message level	🇫🇷 French	View
comparia-votes	User preferences after full conversations	🇫🇷 French	View

Reporting sensitive data

If you find a row in the dataset that you believe contains PII or sensitive data, please let us know via this short form.

Contact

For questions about the data, open a discussion on the dataset discussion page.

---