# LLM Debate Arena

> Structured multi-round debates between AI language models on any topic, with expert personas, fact-checking, and a final judge verdict.

# Context

LLM Debate Arena (llmda.xyz) is a web tool that runs automated debates between large language models via OpenRouter. A user submits a topic; the system assigns expert AI personas, runs N rounds of argumentation, verifies claims with live web search, and produces a final verdict.

Key facts:
- Debate formats: thematic, oxford, socratic, news_digest, hospital_council
- Expert personas: economists, lawyers, scientists, doctors, philosophers, and more
- Fact-checking: claims verified against live DuckDuckGo search results
- Judgement: verdicts are rendered by high-reasoning models (e.g., Google Gemini 2.5 Pro, DeepSeek R1)
- Languages: interface and AI-generated content available in ru, en, de, fr, es, zh, ar, ja
- All published debates are freely readable; each debate exports as clean Markdown via `?format=md`

Debate formats explained:
- thematic: multi-round expert panel discussion (default)
- oxford: formal Oxford-style debate, proposition vs opposition
- socratic: Socratic dialogue exploring underlying assumptions
- news_digest: structured analysis of recent news on the topic
- hospital_council: multi-agent medical consensus for complex clinical decision-making

To find a debate on a specific topic, browse /blog or use the search endpoint in the API.

# Links

- [Blog — all public debates](https://llmda.xyz/blog): browse AI-generated debate transcripts
- [Debate Markdown export](https://llmda.xyz/blog/kak-zavesti-mnogo-podpischikov-v-youtube?format=md): example debate in plain Markdown with YAML frontmatter
- [API documentation](https://llmda.xyz/docs/api): REST API reference
- [Terms of Service](https://llmda.xyz/terms): usage policy