Skip to content
TheChatArchiveThe Chat Archive
Mistralmistral-small-2506· Mistral AIHugging Face LMArena Human Preference 140K

How to download the leaderboard .csv of LMArena from lmarena.ai/leaderboard ?

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-sonnet-4-20250514-thinking-32k’s answer.
  1. User

    How to download the leaderboard .csv of LMArena from lmarena.ai/leaderboard ?

  2. Mistral Assistant

    To download the leaderboard data from LMArena (lmarena.ai/leaderboard) as a CSV file, follow these steps:

    Method 1: Using the Browser (Manual Download)

    1. Open the Leaderboard Page

    2. Inspect the Page (Optional)

      • Right-click on the leaderboard table and select "Inspect" (or press F12/Ctrl+Shift+I).
      • Look for the data source (e.g., an API endpoint or a hidden CSV link).
    3. Check for a Direct Download Link

      • Some websites provide a "Download CSV" button. If available, click it.
      • If not, proceed to the next method.

    Method 2: Using Browser Extensions (If No Direct Download)

    • Use a Web Scraper Extension (e.g., Table Capture for Chrome/Firefox).
      • Install the extension.
      • Right-click the leaderboard table and select "Export to CSV".

    Method 3: Using Python (If API Access is Available)

    If the leaderboard data is loaded via an API, you can fetch it programmatically:

    import requests
    import pandas as pd
    
    # Example (replace with the actual API endpoint)
    url = "https://lmarena.ai/api/leaderboard"
    response = requests.get(url)
    data = response.json()
    
    # Convert to DataFrame and save as CSV
    df = pd.DataFrame(data)
    df.to_csv("lmarena_leaderboard.csv", index=False)
    

    Method 4: Contact Support

    If no direct download option exists, contact LMArena support (via their website or social media) and request the leaderboard data in CSV format.

    Would you like help finding the exact API endpoint if available? Let me know!

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.