Logical Reasoning on PARARULE-Plus SLDNF-diverging 200 instances subset
100Accuracy (All)Claude Formalizer (PrologMCP)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Claude Formalizer (PrologMCP)Model Family=Claude, System Mode=Formalizer2026.06 | 100 | 100 | 100 | 1,548.4 | 6.2 | |
| GPT-4/o4-mini Formalizer (PrologMCP)Model Family=GPT-4/o4-mini, System Mode=Formalizer2026.06 | 99 | 98 | 100 | 1,191.7 | 3.1 | |
| Claude ReasoningModel Family=Claude, System Mode=Reasoning2026.06 | 95 | 91 | 99 | 1,345.4 | 9.5 | |
| GPT-4/o4-mini ReasoningModel Family=GPT-4/o4-mini, System Mode=Reasoning2026.06 | 94 | 94 | 94 | 2,070 | 11.9 |