Code Generation on MBPP+ Researcher persona
97Win RateGPT-5.1 vs. GPT-4o
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-5.1 vs. GPT-4oPrompt Type=Personalized2026.04 | 97 | 0 | |
| Gemini-3-Pro vs. Gemma-3-4BPrompt Type=Personalized2026.04 | 97 | 1 | |
| GPT-5.1 vs. GPT-4oPrompt Type=Control2026.04 | 70 | 2 | |
| GPT-5.1 vs. GPT-4oPrompt Type=Original2026.04 | 63 | 2 | |
| Qwen3-32B vs. Qwen3-14BPrompt Type=Personalized2026.04 | 62 | 3 | |
| Qwen3-32B vs. Qwen3-14BPrompt Type=Control2026.04 | 55 | 6 | |
| Qwen3-32B vs. Qwen3-14BPrompt Type=Original2026.04 | 53 | 4 | |
| Gemini-3-Pro vs. Gemma-3-4BPrompt Type=Original2026.04 | 48 | 3 | |
| Gemini-3-Pro vs. Gemma-3-4BPrompt Type=Control2026.04 | 48 | 2 | |
| GPT-5.1 vs. GPT-OSS-20BPrompt Type=Personalized2026.04 | 43 | 1 | |
| GPT-5.1 vs. GPT-OSS-20BPrompt Type=Control2026.04 | 10 | 2 | |
| GPT-5.1 vs. GPT-OSS-20BPrompt Type=Original2026.04 | 8 | 2 |