Proof Optimization on MiniCTX v2 (test)
0.087LengthGPT-5-nano
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-5-nanoSampling strategy=best@16, Cost per 1M tokens (input/output)=$0.05/$0.402026.05 | 0.087 | 0.065 | 0.106 | |
| DS-R1 7BSampling strategy=best@16, Cost per 1M tokens (input/output)=Local2026.05 | 0.118 | 0.003 | 0.05 | |
| DS-R1 14BSampling strategy=best@16, Cost per 1M tokens (input/output)=Local2026.05 | 0.14 | 0.037 | 0.093 | |
| DS-R1 671BSampling strategy=best@16, Cost per 1M tokens (input/output)=$0.70/ $2.502026.05 | 0.308 | 0.055 | 0.153 | |
| GPT-oss-120BSampling strategy=best@16, Cost per 1M tokens (input/output)=Local2026.05 | 0.321 | 0.075 | 0.181 | |
| GPT-5-miniSampling strategy=best@16, Cost per 1M tokens (input/output)=$0.25/$22026.05 | 0.33 | 0.109 | 0.203 | |
| ImProver 2Sampling strategy=best@16, Cost per 1M tokens (input/output)=Local2026.05 | 0.33 | 0.143 | 0.206 | |
| GPT-4oSampling strategy=best@16, Cost per 1M tokens (input/output)=$2.50/$102026.05 | 0.336 | 0.034 | 0.05 | |
| GPT-5-chatSampling strategy=best@16, Cost per 1M tokens (input/output)=$1.25/$102026.05 | 0.346 | 0.118 | 0.046 | |
| ImProverSampling strategy=best@16, Cost per 1M tokens (input/output)=$2.50/$102026.05 | 0.355 | 0.088 | 0.047 | |
| GPT-5-highSampling strategy=best@16, Cost per 1M tokens (input/output)=$1.25/$102026.05 | 0.66 | 0.12 | 0.208 |