Automated Inequality Proving on PolyIneqBench n=4
40.3Pass RateNSPI
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NSPI2026.05 | 40.3 | 14.4 | |
| Maple2026.05 | 39 | 107.4 | |
| Z32026.05 | 32.5 | 21.8 | |
| LIPS2026.05 | 27.3 | 131 | |
| Gemini-3-Pro2026.05 | 24.7 | 102.4 | |
| GPT-5.22026.05 | 10.4 | 86.7 | |
| DeepSeek-V3.22026.05 | 6.5 | 420.9 | |
| Goedel-Prover-v22026.05 | 5.2 | 136.4 | |
| Kimina-Prover2026.05 | 5.2 | 111.7 | |
| DS-Prover-v22026.05 | 2.6 | 17.9 |