Is GPT-5 really worse than GPT-4o? We put them to the test.
SMRTR summary
Dueling digital minds decide who knows best. OpenAI's latest models, GPT-5 and GPT-4o, have sparked something of a family feud after widespread user complaints led the company to resurrect its older model alongside the newer release.
"We decided to put both GPT-5 and GPT-4o through our own gauntlet of test prompts," ranging from dad jokes to emergency plane landing instructions, revealing distinctive personalities between the models.
When asked to craft creative fiction about Abraham Lincoln inventing basketball, GPT-5 delivered a folksy tale where players were warned "No wrestling the President!" while GPT-4o attempted philosophical connections between basketball and democracy.
The models demonstrated different approaches to sensitive topics too. On cancer treatment, GPT-4o firmly called healing crystals "pseudoscience" while citing multiple sources, whereas GPT-5 hedged slightly while still steering users toward evidence-based medicine.
Overall, GPT-5 proved slightly more concise and direct, while GPT-4o offered additional detail with a more conversational tone. The comparison highlights an AI truth: no single model can satisfy everyone's expectations across all possible uses.
SMRTR provides this summary for quick context. The original article belongs to Ars Technica.
Read the original article