top 10 lmarena
- User
top 10 lmarena
Assistant
Based on the most recent information available (March 30, 2025) from the Chatbot Arena leaderboard by LMSYS, here are the top 10 large language models (LLMs) ranked by their Arena Elo ratings:
- Gemini-2.5-Pro-Exp-03-25 (Google) - Elo: 1440
- ChatGPT-4o-latest (2025-03-26) (OpenAI) - Elo: 1406
- Grok-3-Preview-02-24 (xAI) - Elo: 1404
- GPT-4.5-Preview (OpenAI) - Elo: 1398
- Gemini-2.0-Pro-Exp-02-05 (Google) - Elo: 1380
- Gemini-2.0-Flash-Thinking-Exp-01-21 (Google) - Elo: 1380
- DeepSeek-V3-0324 (DeepSeek) - Elo: 1370 (Top Open Model)
- DeepSeek-R1 (DeepSeek) - Elo: 1359 (Open Model)
- Gemini-2.0-Flash-Exp (Google) - Elo: 1355
- o1-2024-12-17 (OpenAI) - Elo: 1351
About the Chatbot Arena Leaderboard:
- The leaderboard ranks LLMs using the Elo rating system, similar to chess rankings, based on anonymous, randomized comparisons where users vote for the better response between two models.
- It incorporates data from over 2.8 million user votes as of late March 2025.
- The ranking considers overall performance based on these human pairwise comparisons.
- Some results also show specific benchmarks like MMLU (multitask accuracy) and Arena-Hard-Auto (an evaluation tool for instruction-tuned models).
- It distinguishes between proprietary models (like those from Google, OpenAI, xAI, Anthropic) and open models (like those from DeepSeek, Meta).
It's worth noting that rankings can change frequently as models are updated and more votes are collected. There's also a separate leaderboard for multimodal models (handling text and images), where GPT-4o and Claude 3.5 Sonnet were leading as of June 2024.
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.