Benchmarks · creative
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).
30 models · 30 results · data as of 2026-10-29 · eqbench.com ↗
Leaderboard
Opus 5 max
All results
30 rows| # | Model | Effort | EQ-Bench 4 |
|---|---|---|---|
| 1 | Opus 5 | max | 1360 |
| 2 | Opus 5.5 | max | 1339 |
| 3 | Kimi K3 | max | 1318 |
| 4 | Fable 5 | max | 1310 |
| 5 | Opus 4.7 | max | 1294 |
| 6 | GPT-5.5 | xhigh | 1294 |
| 7 | Opus 4.8 | max | 1267 |
| 8 | GPT-5.4 | xhigh | 1260 |
| 9 | Muse Spark 1.1 | max | 1252 |
| 10 | Gemini 4 Argon | max | 1240 |
| 11 | GPT-5.6 Sol | max | 1240 |
| 12 | Sonnet 5 | max | 1225 |
| 13 | GPT-5.6 Terra | max | 1224 |
| 14 | Inkling | max | 1216 |
| 15 | Opus 4.6 | max | 1213 |
| 16 | GLM 5.2 | max | 1211 |
| 17 | MiMo V2.5 Pro | xhigh | 1201 |
| 18 | Sonnet 4.6 | max | 1200 |
| 19 | Kimi K2.6 | max | 1194 |
| 20 | DeepSeek V4 Pro | max | 1160 |
| 21 | GPT-5.6 Luna | max | 1149 |
| 22 | MiniMax M3 | max | 1144 |
| 23 | Gemini 3.1 Pro | high | 1136 |
| 24 | Gemma 4 31B IT | high | 1114 |
| 25 | Qwen 3.7 Max | max | 1104 |
| 26 | Gemini 3.5 Flash | xhigh | 1081 |
| 27 | Grok 4.3 | xhigh | 1069 |
| 28 | Haiku 4.5 | xhigh | 1057 |
| 29 | Qwen 3.6 27B | max | 1019 |
| 30 | Mistral Medium 3.5 | high | 988 |