This content originally appeared on DEV Community and was authored by Emir Taner
OpenAI is softening GPT-5’s tone after a rocky launch. The update promises subtle warmth without losing precision — but will it really improve the experience, or just add fluff?
OpenAI just announced that it’s making GPT-5 a little more “human.” After a launch that even CEO Sam Altman admitted was “a little more bumpy than we’d hoped for,” the company is rolling out an update aimed at making the model warmer and friendlier.
Some early users had complained that GPT-5 felt too blunt compared to GPT-4o. OpenAI says the change is subtle but intentional: you’ll now notice small touches like “Good question” or “Great start” sprinkled into responses — designed to make the AI feel more approachable without slipping into unnecessary flattery.
VP Nick Turley put it simply: the previous GPT-5 personality was “very to the point,” but now the model should come across less robotic and a bit more natural in conversation.
My Take
I can see why OpenAI is taking this step. Users don’t just care about accuracy — they care about tone. If your tool feels cold or dismissive, people disengage faster, no matter how powerful it is under the hood.
At the same time, I’m cautious. Personally, I value clarity and precision over friendliness. If GPT-5 becomes too “nice” and less direct, it could feel slower to get to the point — which is frustrating when you’re in a flow state, coding, or debugging. Ideally, this friendliness should remain a layer, not the foundation.
Final Thought
I’ll be testing the update over the next week to see if it really makes conversations more natural or if it just adds fluff. For devs, product builders, and power users, balance is key: we want tools that are approachable and razor-sharp.
This content originally appeared on DEV Community and was authored by Emir Taner