AI Intelligence
Model performance on controlled practical tasks.
- 1,224 task results
- 3 model families
- 17 historical configurations
Study files
Download the task definitions, or open Records for measured runs and their original evidence.
This download defines Core 1.1.0. Available measurements use Core 1.0.7; each record retains that version.
Current benchmark versions
Task scorer, aggregate scoring, and configuration design
Current benchmark versions
Task scorer, aggregate scoring, and configuration design- Task set
- 1.1.0 · 72 tasks in 10 domains
- Task scorer
- 1.0.6
- Aggregate scoring
- 1.0.8
- Configuration design
- 17 configurations; design count, not completed runs