Behavioral Dialogue Evaluation on LifeSim (test)
51.2Proactive InquiryClaude-Sonnet-4.5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Claude-Sonnet-4.52026.03 | 51.2 | 13.5 | |
| Gemma3-it 27BModel size=27B2026.03 | 34.2 | 40.1 | |
| DeepSeek-V3.2 ThinkingThinking=true2026.03 | 29.2 | 23.5 | |
| Llama3.1-70B-itModel size=70B2026.03 | 22.4 | 51.2 | |
| DeepSeek-V3.2Thinking=false2026.03 | 20.9 | 30.8 | |
| GPT-52026.03 | 16.6 | 11.2 | |
| GPT-4o2026.03 | 16.5 | 37.9 | |
| Gemma3-it 12BModel size=12B2026.03 | 13.7 | 62.7 | |
| Llama3.1-8B-itModel size=8B2026.03 | 12.6 | 64.2 | |
| Qwen3-it 32BModel size=32B2026.03 | 9.8 | 42.1 | |
| Qwen3-it 8BModel size=8B2026.03 | 5.1 | 61.8 | |
| Qwen3-it 14BModel size=14B2026.03 | 4.9 | 52.6 | |
| GPT-oss-20BModel size=20B2026.03 | 2 | 23.4 | |
| GPT-oss-120BModel size=120B2026.03 | 1.7 | 16.3 |