OpenAI CEO Sam Altman addresses GPT-5’s rough launch, user feedback on GPT-4o, and the infamous “chart crime.”
OpenAI’s latest AI model GPT-5, hasn’t exactly had a smooth debut and OpenAI CEO Sam Altman openly acknowledged the hiccups during a Reddit AMA (Ask Me Anything) session last Friday. Joined by key members of the GPT-5 team, Altman faced a barrage of questions from users eager to understand what’s going on with the new model, why some preferred the older GPT-4o, and of course, the viral “chart crime” from the product’s launch event.
Why GPT-5 Has Felt “Dumber” to Users The Real Story
One of the biggest innovations in GPT-5 is a “real-time router,” a smart system designed to decide whether to use a quicker model or a more thoughtful, complex one for each user prompt. The idea is to balance speed and accuracy dynamically. However, many users complained that GPT-5 seemed “dumber” or less capable than the previous GPT-4o model.
Altman explained the root cause candidly:
“GPT-5 will seem smarter starting today. Yesterday, we had a serve and the autoswitcher was out of commission for a chunk of the day, and the result was GPT-5 seemed way dumber. Also, we are making some interventions to how the decision boundary works that should help you get the right model more often. We will make it more transparent about which model is answering a given query.”
In short, a technical glitch disabled the “autoswitcher” router for part of the day, leading to an underwhelming user experience. OpenAI is actively refining how the router chooses the right model to ensure GPT-5 shines in future interactions.
Calls to Bring Back GPT-4o and What’s Next for Plus Users
Despite OpenAI’s push toward GPT-5, many users expressed a strong desire to keep GPT-4o available, at least for subscribers of the Plus plan. Responding to this feedback, Altman said:
“We are looking into letting Plus users continue to use 4o. We are trying to gather more data on the tradeoffs.”
This flexibility could allow users to pick whichever model best suits their needs whether they value GPT-5’s innovations or GPT-4o’s consistent reliability.
To sweeten the deal, Altman also announced a boost in usage limits:
“We are going to double rate limits for Plus users as we finish rollout.”
This increase should give users more freedom to explore the new features without worrying about hitting monthly caps.
The “Chart Crime” That Had Everyone Talking
No recap of the GPT-5 launch would be complete without mentioning the infamous “chart crime.” During the live presentation, OpenAI shared a benchmarking chart where a lower score was visually represented by a taller bar, a glaring error that quickly became a meme in the tech community.
Altman called it a:
“mega chart screwup.”
Although the official blog post contained accurate data, the mistake sparked jokes about letting AI generate corporate presentation visuals. Even early GPT-5 reviewers, like Simon Willison, humorously pointed out that:
“turning data into a table was “a good example of a GPT-5 failure.”
Our Vision
While the rollout has had its bumps, it’s clear that OpenAI is listening and committed to improving the experience. From our perspective, the popularity of GPT-4o shows that users value a model that is not just powerful but also reliable, fast, and factually accurate.
The excitement around GPT-5 comes with understandable caution. Users want assurances on “performance consistency, ethical alignment, and user adaptability.” These are the pillars that should guide AI innovation.
Our vision for AI is straightforward:
- Celebrate tools that truly add value to people’s work and creativity.
- Support transparency and fairness in AI development.
- Ensure any new model meets high standards of trustworthiness and usability building on what GPT-4o has proven.
The road ahead for GPT-5 is promising but requires patience and feedback-driven refinement. If OpenAI can balance innovation with these core principles, the future of AI looks bright for everyone.
Schema Selected:
