Boolean Satisfiability (SAT) on SAT Hard (test)
64AccuracyRecursive Model
Evaluation Results
| Method | Links | |
|---|---|---|
| Recursive ModelBackbone=Qwen2.5-3B-Instruct, Framework=Recursive2026.03 | 64 | |
| LLaMA3.3-70B2026.03 | 52.9 | |
| Qwen3-235B2026.03 | 51.4 | |
| Random Baseline2026.03 | 50 | |
| GPT-4o2026.03 | 48.8 | |
| DeepSeek-Distill-14B2026.03 | 46.4 |