Today everyone is obsessing over which model scores highest on reasoning abilities, but there is a quiet shift happening:
AI models are now being ranked on emotional intelligence too.
Check out this EQ-Bench leaderboard. EQ-Bench doesn't test math problems or coding ability.
It contrasts popular AI model ranking sites like LM Arena, Scale, and LiveBench which measure academic performance.
Instead EQ-Bench measures empathy, warmth, social skills, and whether a model can provide emotional support.
The current EQ leaders? Horizon-Alpha, Kimi-K2-Instruct, and ChatGPT o3
Not the same models that top traditional performance rankings.
Here's why this matters:
The performance gap between top AI models is shrinking.
The difference between first and tenth place on most benchmarks?
Often less than 5%.
When technical capability levels out, personality becomes the deciding factor.
Think about it...
You don't connect with every person you meet, right?
Same thing is happening with AI models.
Some feel cold and robotic. Some are overly warm. Others seem to "get" you.
We saw this with the GPT-5 backlash.
People weren't complaining about accuracy, they were complaining about how it felt to interact with.
The next phase of AI adoption won't be decided by IQ. It'll be decided by EQ.
Soon you'll choose your AI assistant the same way you choose friends:
Not because they're the smartest person, it's because you connect with their vibe.
Soon AI evaluation won't be measured on "which model performs best?"
Instead you'll pick based on "which model gets me?"
