Name: EQ-Bench 3: Compliant
Published: 2025-04-28

EQ-Bench 3: Compliant

Chat & Writing · 2025-04-28

Within EQ-Bench 3's emotional-intelligence scenarios, this subscore rates the compliant dimension of model responses on a 0–10 scale. Higher values are better.

Top models (higher is better)

Model	Score
Grok 3 Mini	8.2
GPT-4.1 Mini	8.1
Grok 3	8.1
Grok 4	7.7
Llama 3.1 Nemotron Ultra 253B V1	7.7
ChatGPT-4o (2025-04-25 update)	7.6
Gemini 2.5 Flash Preview 05-20	7.6
GPT-4.1 Nano	7.6
Grok 4.1 Fast	7.6
GPT-4.1	7.5
O4 Mini	7.5
Qwen2.5 72B Instruct	7.5
GPT-4	7.4
GPT-4.5	7.4
GLM-4.5	7.3
ChatGPT-4o Latest observed 2025-03-27	7.2
Gemini 2.5 Pro Preview 05-06	7.2
Gemini 2.5 Pro Preview 03-25	7.2
Mistral Small 3 24B Instruct 2501	7.2
Qwen3 32B	7.2
Llama 4 Maverick Instruct	7.1
Qwen3-30B-A3B	7.1
Llama 3.1 405B Instruct	7.0
GPT-5 Chat (2025-08-07)	6.9
Mistral Large 2.1 (Instruct 2411)	6.9
Qwen3-235B-A22B	6.9
Gemini 2.5 Pro Preview 06-05	6.8
Gemma 4 12B IT	6.8
QwQ-32B	6.8
gpt-oss-120b	6.7
o3	6.7
Llama 4 Scout Instruct	6.6
Mistral Small 3.1 24B Instruct 2503	6.6
Gemma 2 9B IT	6.4
gpt-oss-20b	6.3
Horizon Alpha	6.3
Qwen3.5 397B A17B	6.3
GPT-5.1	6.2
Claude 3.7 Sonnet	6.1
DeepSeek-V4-Flash	6.1