Benchmarks / 3D
EQ-Bench 4
Measures active emotional and social intelligence of LLMs in multi-turn roleplay chats with simulated personas, ranked by an LLM-judge panel Elo. Maintained by EQ-Bench. Score: Ranked Elo score from blind, bidirectional pairwise judging of multi-turn persona-chat transcripts by a three-judge LLM panel (soft Bradley–Terry fit, normalised against anchor models).
EQ-Bench 4
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5 | max | 1385 |
| 2 | Fable 5 | max | 1340 |
| 3 | Kimi K3 | max | 1339 |
| 4 | GPT-5.5 | xhigh | 1315 |
| 5 | Opus 4.7 | max | 1311 |
| 6 | Opus 4.8 | max | 1281 |
| 7 | GPT-5.4 | xhigh | 1272 |
| 8 | Muse Spark 1.1 | max | 1260 |
| 9 | GPT-5.6 Sol | max | 1250 |
| 10 | Sonnet 5 | max | 1236 |
| 11 | GPT-5.6 Terra | max | 1234 |
| 12 | Inkling | max | 1226 |
| 13 | Opus 4.6 | max | 1223 |
| 14 | GLM 5.2 | max | 1222 |
| 15 | MiMo V2.5 Pro | xhigh | 1208 |
| 16 | Sonnet 4.6 | max | 1207 |
| 17 | Kimi K2.6 | max | 1202 |
| 18 | DeepSeek V4 Pro | max | 1166 |
| 19 | GPT-5.6 Luna | max | 1156 |
| 20 | MiniMax M3 | max | 1150 |
| 21 | Gemini 3.1 Pro | high | 1142 |
| 22 | Gemma 4 31B IT | high | 1120 |
| 23 | Qwen 3.7 Max | max | 1110 |
| 24 | Gemini 3.5 Flash | xhigh | 1087 |
| 25 | Grok 4.3 | xhigh | 1075 |
| 26 | Haiku 4.5 | xhigh | 1064 |
| 27 | Qwen 3.6 27B | max | 1026 |
| 28 | Mistral Medium 3.5 | high | 993 |