Chess
Standard chess, given as FEN. Stockfish 19 scores every legal move by expected score; the answer is a UCI move. See Methodology.
Quality is the mean oracle score of the moves played, × 100. The best move scores 100; an illegal move or no move scores 0. Protocol boardbench-0.2.
Quality vs cost
7 of 7 models
- Anthropic
- OpenAI
- DeepSeek
- Z.ai
- Meta
- Pareto frontier: nothing plotted is both cheaper and better
- Anthropic
- OpenAI
- DeepSeek
- Z.ai
- Meta
Horizontal axis is logarithmic.
Table
| Model | Quality | Legality | Cost | n |
|---|---|---|---|---|
| Opus 5.5High | 33.3 | 100.0 | $0.08 | 3 |
| DeepSeek V4.1 FlashHigh | 33.3 | 100.0 | $0.04 | 3 |
| GLM-5.3 FlashHigh | 33.3 | 100.0 | $0.01 | 3 |
| SolHigh | 33.3 | 100.0 | $0.10 | 3 |
| Haiku 5.5High | 0.0 | 66.7 | <$0.01 | 3 |
| LunaHigh | 0.0 | 100.0 | <$0.01 | 3 |
| Muse Spark 1.3High | 0.0 | 33.3 | $0.11 | 3 |
Headline is the mean move score × 100. n is answers: puzzles × samples.