Open Data

Comparisons on Tenet are public by default. We believe open data makes AI evaluation better for everyone.

Public by Default

When you submit a comparison, your prompt and all AI responses are public by default and appear in the feed. You can toggle "Share to feed" on any comparison you own to remove it from the feed and future exports. Copies already downloaded cannot be recalled.

Why Open?

  • Reproducibility — Anyone can verify our leaderboard results by examining the underlying comparisons.
  • Research — Researchers can study how different models respond to the same prompts across many categories.
  • Transparency — No cherry-picking. Every comparison counts, good or bad.
  • Community — The dataset grows with every comparison, making it more valuable over time.

What's Included

Each record in the dataset contains:

  • The original prompt
  • Each model's complete response
  • Which models were compared
  • Vote tallies (which model users preferred)
  • Timestamp of the comparison

What's NOT Included

We deliberately exclude:

  • User IDs or email addresses
  • API costs or pricing information
  • Any personally identifiable information

Export API

Access the complete dataset programmatically:

# JSON format (paginated)

GET https://tenet-ochre.vercel.app/api/export?format=json&limit=100&offset=0

# JSONL format (streaming, one record per line)

GET https://tenet-ochre.vercel.app/api/export?format=jsonl&limit=1000

Parameters

formatjson or jsonl (default: json)
limitRecords per page, max 1000 (default: 100)
offsetSkip this many records (default: 0)
sinceUnix timestamp (ms) — only records after this time

Response Schema

{
  "data": [
    {
      "id": "abc123",
      "prompt": "What is 2+2?",
      "createdAt": 1698765432000,
      "responses": [
        {
          "modelId": "openai/gpt-4",
          "modelName": "GPT-4",
          "content": "2+2 equals 4."
        },
        {
          "modelId": "anthropic/claude-3",
          "modelName": "Claude 3",
          "content": "The answer is 4."
        }
      ],
      "votes": {
        "openai/gpt-4": 5,
        "anthropic/claude-3": 3
      }
    }
  ],
  "pagination": {
    "total": 1234,
    "limit": 100,
    "offset": 0,
    "hasMore": true
  }
}

License

The Tenet dataset is released under Creative Commons Attribution 4.0 International (CC BY 4.0).

You're free to use the data for any purpose, including commercial use, as long as you give appropriate credit. A simple "Data from Tenet" with a link back is sufficient.

Before You Compare

Because data is public by default, don't submit:

  • Personal information (yours or anyone else's)
  • Confidential business data
  • Proprietary code or documents
  • Anything you wouldn't want on the public internet

Questions about the dataset? Read our terms or privacy policy.