EIBench (Emotion Management) - Human Arena: leaderboard
The human arena of EIBench (2026), a benchmark for interactive emotion management: annotators converse with two anonymized models under the same scenario profile and choose which handles it better. The board is an Elo rating fitted from more than 4,000 such pairwise judgments; higher is better.
Metric: Elo rating from more than 4,000 human pairwise judgments: annotators converse with two anonymized models under the same EIBench scenario profile and pick the one that handles it better; higher is better. Source: arxiv.org. Saturation forecast: Around June 2027. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 1616 | #94 |
| 2 | Gemini 3 Flash | 1606 | #164 |
| 3 | GPT-5.4 | 1593 | #123 |
| 4 | Claude Sonnet 4.6 | 1548 | #161 |
| 5 | Seed 2.0 Pro | 1536 | #211 |
| 6 | Qwen 3.6 Max Preview | 1519 | #159 |
| 7 | GLM-5.1 | 1504 | #179 |
| 8 | Qwen 3 Max | 1487 | #403 |
| 9 | DeepSeek V4 Pro | 1484 | #188 |
| 10 | Kimi K2.6 | 1478 | #178 |
| 11 | Grok 4.20 | 1476 | #297 |
| 12 | MiniMax-M2.5 | 1471 | #582 |
| 13 | Grok 4.20 (Reasoning) | 1458 | #297 (Grok 4.20) |
| 14 | Qwen 3 32B | 1405 | #960 |
| 15 | Qwen 3 8B | 1316 | #1830 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=eibench-emotion-management-human-arena · How It Works · Data refreshed daily, snapshot 2026-11-13.