Preference Elicitation on SpecBench Task 1
74Improvement Rate (T0->T5)Buddy
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Buddy2026.05 | 74 | 77 | 92 | — | — | 0 | |
| Claude CodeBackbone=Opus-4.62026.05 | 62 | 67 | 80 | — | — | 1 | |
| Cursor CLIBackbone=GPT-5.4-mini2026.05 | 62 | 61 | 78 | — | — | 3 | |
| Gemini CLIBackbone=Gemini-2.5-pro2026.05 | 50 | 57 | 61 | — | — | 17 |