Instruction Following on Inverse IFEval, MARS-Bench, MultiChallenge, Collie-Hard, EIFBench
99Pass@1GPT-5 High
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5 HighDataset=Collie-Hard2026.03 | 99 | |
| Gemini 3-proDataset=Collie-Hard2026.03 | 95.5 | |
| Gemini 3-proDataset=MARS-Bench2026.03 | 80.8 | |
| Gemini 3-proDataset=Inverse IFEval2026.03 | 80.6 | |
| Seed1.8Dataset=Inverse IFEval2026.03 | 80.3 | |
| GPT-5 HighDataset=Inverse IFEval2026.03 | 78.9 | |
| Claude Sonnet-4.5Dataset=Collie-Hard2026.03 | 77.6 | |
| GPT-5 HighDataset=MARS-Bench2026.03 | 77.2 | |
| Gemini 2.5-proDataset=Inverse IFEval2026.03 | 75.3 | |
| Gemini 2.5-proDataset=MARS-Bench2026.03 | 73.6 | |
| Seed1.8Dataset=Collie-Hard2026.03 | 72.6 | |
| Claude Sonnet-4.5Dataset=MARS-Bench2026.03 | 72.5 | |
| Claude Sonnet-4.5Dataset=Inverse IFEval2026.03 | 70.2 | |
| Seed1.8Dataset=MARS-Bench2026.03 | 70.1 | |
| GPT-5 HighDataset=MultiChallenge2026.03 | 69.6 | |
| Gemini 2.5-proDataset=Collie-Hard2026.03 | 69.5 | |
| Gemini 3-proDataset=MultiChallenge2026.03 | 67.4 | |
| Seed1.8Dataset=MultiChallenge2026.03 | 66.7 | |
| GPT-5 HighDataset=EIFBench2026.03 | 66.7 | |
| Claude Sonnet-4.5Dataset=MultiChallenge2026.03 | 57.2 | |
| Gemini 2.5-proDataset=MultiChallenge2026.03 | 55.4 | |
| Gemini 3-proDataset=EIFBench2026.03 | 50.1 | |
| Seed1.8Dataset=EIFBench2026.03 | 48.6 | |
| Claude Sonnet-4.5Dataset=EIFBench2026.03 | 47 | |
| Gemini 2.5-proDataset=EIFBench2026.03 | 44.7 |