Suite coverage
ELO ladder evaluation against classic engines, plus tactical puzzle sets and positional endgame studies.
OmniLink
Build an Agent
Benchmark
Chess
Strategic planning benchmarks that evaluate search depth, memory, and move-by-move compositional reasoning.
Metrics monitored
- Blunder rate and best-move agreement.
- Long-term plan consistency across 30+ move horizons.
- Adaptation to novel openings and mid-game disruptions.
Current focus
Sharpening endgame conversion rates while keeping explainability accessible to human analysts.