The launch of GPT-5 hasn’t gone as smoothly as many expected. Users and developers have reported several issues ranging from technical faults like broken auto model-switching to slower and less consistent responses via the API. On top of that, many in the community say GPT-5 feels like a step backward compared to GPT-4o citing shorter, less creative replies and a bland tone.

The decision to limit ChatGPT Plus users to 200 messages per week hasn’t helped, causing frustration among paying customers. Adding fuel to the fire, the removal of previous model options left users feeling boxed in.

There was also that infamous “crime graph” incident during the GPT-5 presentation a misleading performance chart that backfired on OpenAI’s messaging and led to plenty of online ridicule.

In response, OpenAI is working on restoring access to GPT-4o, doubling message limits, and improving transparency around model selection. Despite GPT-5’s strengths in coding and image generation, the mixed reception so far highlights the challenges of pushing the envelope while keeping user trust intact.

What’s your take? Is GPT-5 a genuine leap forward, or has OpenAI stumbled with this release?