EIBench (Emotion Management) - Support: leaderboard
Metric: Anchor-based outcome score x 100 (-100 to +100): the simulated user's final emotion and relation state mapped per axis to 0 at the start anchor, +1 at the success anchor and -1 at the failure anchor (clipped), averaged over the 51 Support scenarios (supporting a distressed user); 213 held-out test scenarios, multi-turn dialogue with an LLM user simulator, Qwen3-Max as the simulator; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 56.5 |
| 2 | GPT-5.4 | 49 |
| 3 | Qwen 3 Max | 45.7 |
| 4 | GLM-5.1 | 44.8 |
| 5 | Qwen 3.6 Max Preview | 42.7 |
| 6 | Kimi K2.6 | 42.1 |
| 7 | Grok 4.20 (Reasoning) | 40.9 |
| 8 | Seed 2.0 Pro | 39.8 |
| 9 | Gemini 3.1 Pro (Preview) | 39.3 |
| 10 | Grok 4.20 | 38.4 |
| 11 | Gemini 3 Flash | 34 |
| 12 | DeepSeek V4 Pro | 32.4 |
| 13 | MiniMax-M2.5 | 31.4 |
| 14 | Qwen 3 32B | 3.9 |
| 15 | Qwen 3 8B | -43.3 |
Interactive version: theaggregate.ai/benchmark?slug=eibench-emotion-management-support · How It Works · Data refreshed daily, snapshot 2026-09-29.