EQ-Bench 3: leaderboard
LLM-judged emotional intelligence benchmark using multi-turn role-plays involving relationship conflicts and workplace dilemmas to measure empathy, social skills, and psychological insight.
Metric: Elo. Source: eqbench.com. Saturation forecast: Around April 2027. 75 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 1460.9 | #49 |
| 2 | Claude Sonnet 4.6 | 1384.2 | #90 |
| 3 | Claude Opus 4.6 | 1383.4 | #63 |
| 4 | DeepSeek V4 Pro | 1329.7 | #102 |
| 5 | GPT-5.4 | 1328 | #79 |
| 6 | GPT-5.5 | 1327.9 | #31 |
| 7 | Kimi K2.6 | 1319.2 | #104 |
| 8 | GPT-5.1 (2025-11-13) | 1314.4 | #101 |
| 9 | GPT-5.2 | 1314.2 | #110 |
| 10 | Kimi K2 | 1311 | #243 |
| 11 | GLM-5.1 | 1309.6 | #97 |
| 12 | Claude Opus 4.5 (20251101) | 1308.4 | #74 |
| 13 | GLM-5 | 1307.3 | #141 |
| 14 | Gemini 3 Pro (Preview) | 1307.1 | #66 |
| 15 | Gemini 2.5 Pro (Preview 06-05) | 1304.7 | #106 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=eq-bench-3 · How It Works · Data refreshed daily, snapshot 2026-10-20.