When OpenAI dropped GPT-5 last week, the big sell was a streamlined ChatGPT experience. The company envisioned a single, versatile model smart enough to pick the best way to answer any query automatically. No more digging through a complicated model picker — a menu OpenAI CEO Sam Altman himself admitted he found frustrating.
But guess what? That tidy setup was short-lived. Almost immediately, the model picker made a comeback, now with even more choices than before. So what happened? And what does this tell us about how people actually want to interact with AI today?
A unified model with a twist: The plan and its challenges
OpenAI‘s original idea was bold: build GPT-5 with a built-in “router” that would decide in a snap whether to prioritize speed, depth, or tone for every user question. This way, users wouldn’t have to choose manually, the system would do it for them. Many welcomed the simplicity, especially those overwhelmed by the dizzying array of previous models.
But GPT-5’s launch wasn’t entirely smooth sailing. On day one, the routing system stumbled, delivering slower or sometimes less sharp responses than users expected. Behind the scenes, OpenAI’s leadership made it clear they saw this as a first draft – a starting point to improve quickly. The technology needed to analyze the question’s nature and the user’s expectations and then pick the right AI model in milliseconds. Not an easy feat!
What’s clear is that a one-size-fits-all approach didn’t fully capture how people want to engage with AI. Some users like quick, snappy answers, others prefer thorough deep dives, and many develop emotional attachments to specific AI personalities.
New GPT-5 modes and the return of the classics
Responding to user feedback, OpenAI rolled out new modes for GPT-5:
- Auto – the default smart routing mode that tries to pick the best response style automatically.
- Fast – designed for quick and concise replies when you’re in a hurry.
- Thinking – slower but more detailed answers for when depth really matters.
But that’s not all. Paid ChatGPT users can now once again choose from several beloved legacy models like GPT-4o, GPT-4.1, and o3. In fact, GPT-4o made a default comeback in the picker, acknowledging how many people preferred its warmer, friendlier personality compared to GPT-5’s more neutral tone.
Altman openly shared that OpenAI is working on updating GPT-5’s personality to be warmer but without the sometimes polarizing quirks of GPT-4o. Even more interesting: plans for per-user personality customization are on the horizon. Imagine tuning your AI’s style to fit your own vibe — from formal and analytical to casual and chatty.
OpenAI’s latest learning? “We really just need to get to a world with more per-user customization of model personality.”
What this means in the bigger picture
The GPT-5 rollout and subsequent tweak highlight several fascinating dynamics at play in AI’s evolving relationship with users. First, users aren’t just looking for technical capability; they want personalized experiences that feel intuitive and even relatable. The emotional connection to AI models is real, there were even public “funerals” for discontinued bots like Anthropic’s Claude 3.5 Sonnet, showing how people project personality and form attachments.
Second, it’s a reminder that AI development is iterative. No launch is perfect. OpenAI’s transparency about struggles and rapid updates is a positive sign. Balancing speed, accuracy, and personality in real time to millions of users is a monumental challenge.
Finally, this pivot back to giving users more explicit choices reflects a broader trend: control and customization matter. People want flexibility in how AI serves them, not a one-size-fits-all magic bullet.
Key takeaways for ChatGPT users and AI enthusiasts
- Try the new GPT-5 modes to see what fits your pace and need – Auto for balanced, Fast for quick chats, or Thinking for thoughtful responses.
- If you loved GPT-4o or other favorite models, you’re in luck, they’re back for paid users and offer a different tone and style worth exploring.
- Keep an eye out for more personality customization features in future updates, soon you might tailor not just content but the way ChatGPT feels to you.
OpenAI’s dance between simplicity and complexity with GPT-5 reminds us that AI isn’t just about raw power. It’s also about crafting experiences that resonate with how real people think, feel, and want to interact. I find it exciting to watch this technology evolve so openly and responsively, it’s like witnessing the AI world learn to become more human, one tweak at a time.
