All open datasets
Explore open, documented datasets for analysis and machine learning, with clear licences, JSON API access, and Parquet downloads.
4 datasets
Every dataset here is published as immutable Apache Parquet versions with a documented column schema, column profiles, an explicit licence and a named publisher. Read the schema and a sample of real rows without an account; sign up free to run bounded JSON queries or download the whole snapshot for training, backtesting or analysis. Agents can do the same over the Model Context Protocol at /api/mcp, or read /llms-full.txt for the whole catalogue in one file.
- Daily
Mostly Right5 Nov 2026, 19:24:47 UTCLLM benchmark scores by model, since 2019
- 7.3K rows (estimated)
- 2,196 views
- 105 uses. Uses include downloads and successful API/MCP queries
- DailyMostly Right24 Sept 2026, 11:01:28 UTC
Y Combinator company status and Form D filings
- 9.9K rows (estimated)
- 1,285 views
- 180 uses. Uses include downloads and successful API/MCP queries
- Mostly Right9 Oct 2026, 03:02:00 UTC
Free Steam games with descriptions and screenshots
- 23.1K rows (estimated)
- 916 views
- 134 uses. Uses include downloads and successful API/MCP queries
- Mostly Right30 Oct 2026, 06:04:32 UTC
Every Hugging Face model classified by what it does
- 885.4K rows (estimated)
- 556 views
- 41 uses. Uses include downloads and successful API/MCP queries
