Atlas

Benchmarks

← All benchmarks

EQ-Bench 3: Assertive

Chat & Writing · 2025-04-28

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

Top models (higher is better)

ModelScore
Kimi K2.57.7
Sonnet 4.57.6
GPT-5.57.6
Kimi K2.67.6
Gemini 3 Pro Preview7.5
GPT-5.27.5
Opus 4.67.3
Gemma 4 31B IT7.3
GPT-5.47.3
Grok 4.207.3
Claude Fable 57.1
Opus 4.77.1
GLM-5.27.1
GLM-5.17.0
Opus 4.86.9
Nanbeige4-3B-Thinking-25116.8
Gemma 4 26B A4B IT6.7
Sonnet 46.6
Gemini 3.1 Pro Preview6.6
GLM 4.7 Flash6.6
Hivemind-32B-Preview6.6
Sonnet 4.66.5
GLM-4.76.5
GPT-5.16.5
Qwen3.5 397B A17B6.5
Opus 4.56.4
GLM-56.4
Kimi K2 Instruct6.4
Horizon Alpha6.3
Hermes 4 405B6.2
DeepSeek-V4-Pro6.0
Opus 45.8
DeepSeek-V4-Flash5.8
Mistral Small 3.2 24B Instruct 25065.8
DeepSeek-R15.7
Gemma 3 27B IT5.7
Claude 3.5 Sonnet (Oct 2024)5.6
Claude 3.7 Sonnet5.5
o35.5
Gemma 4 12B IT5.3
Loading Atlas data…