Auto-formalization on ProofNet (test)
97.9Pass@8LongCat-Flash-Prover
Evaluation results
| Method | Links | |
|---|---|---|
| LongCat-Flash-Proverw/ TIR=true2026.03 | 97.9 | |
| Gemini-3 Pro2026.03 | 91.9 | |
| Claude-Opus-4.52026.03 | 90.9 | |
| Kimi-K2.52026.03 | 88.2 | |
| LongCat-Flash-Prover2026.03 | 87.1 | |
| DeepSeek-V3.22026.03 | 81.8 | |
| Goedel-V2-Formalizer-32B2026.03 | 79 | |
| Goedel-V2-Formalizer-8B2026.03 | 76.9 | |
| StepFun-Formalizer-32B2026.03 | 62.4 | |
| Kimina-Autoformalizer-7B2026.03 | 55.4 | |
| StepFun-Formalizer-7B2026.03 | 53.8 | |
| ATF-32B2026.03 | 53.8 | |
| ATF-8B-Distilled2026.03 | 45.7 |