Code Generation on Hard Corpus 192 problems 4
83.3Correctness (pass@1)AlgoSkill
Evaluation Results
| Method | Links | |
|---|---|---|
| AlgoSkillBackbone=GPT-5.52026.06 | 83.3 | |
| DirectBackbone=GPT-5.52026.06 | 81.8 | |
| CoTBackbone=Sonnet-4-52026.06 | 77.1 | |
| AlgoSkillBackbone=Sonnet-4-52026.06 | 77.1 | |
| CoTBackbone=GPT-5.52026.06 | 76.6 | |
| DirectBackbone=Sonnet-4-52026.06 | 75 | |
| AlgoSkillBackbone=gpt-oss-120b2026.06 | 75 | |
| AlgoSkillBackbone=Haiku2026.06 | 71.4 | |
| DirectBackbone=Haiku2026.06 | 70.3 | |
| CoTBackbone=Haiku2026.06 | 67.7 | |
| DirectBackbone=gpt-oss-120b2026.06 | 54.7 | |
| CoTBackbone=gpt-oss-120b2026.06 | 52.6 |