what are online LLM in lmarena
- User
what are online LLM in lmarena
Assistant
LMArena's Chatbot Arena evaluates a diverse range of online LLMs through live, community-driven comparisons. The platform focuses on publicly available models, including both commercial and open-source options. Key models tested include:
Commercial Models
- GPT-4 (via API)
- Gemini/Bard (Google's models)
- Claude (Anthropic's models, including Claude 3.5 Sonnet)
- DeepSeek-R1 (DeepSeek's model)
Open-Source Models
- Llama (Meta's model)
- Mistral (Mistral AI's model)
- Vicuna (LMSYS's fine-tuned LLaMA variant)
- Alpaca (Stanford's instruction-following model)
- ChatGLM (Tsinghua University's bilingual model)
Specialized Evaluations
The WebDev Arena subset tests coding capabilities, featuring models like:
- GPT-4
- Gemini
- DeepSeek
- Qwen
The platform uses an Elo rating system and Bradley-Terry model to rank models dynamically based on user votes[1][2][4][7]. Models remain on the leaderboard for at least two weeks after stabilization[2]. For real-time rankings, users can visit the live leaderboard[6][1][7].
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.