First Impressions of GPT-5

GPT-5 arrived with an enormous amount of expectation attached to it. By the time it launched, OpenAI had already set a precedent with GPT-4, GPT-4o, and the reasoning models that followed. Each release raised the bar, not just technically, but emotionally. As users, we have come to expect clear leaps forward, and GPT-5 seemed to miss the high benchmark mark for the majority of users.

From a purely technical standpoint, GPT-5 is undeniably powerful. It improves on reasoning, long-context handling, and multimodal capability. Yet the initial reaction across the AI community was mixed, and in some cases openly critical. That contrast between capability and perception is what makes GPT-5 such an interesting release to reflect on.

The Weight of Expectation

By the time GPT-5 was announced, expectations were no longer reasonable. They were enormous. GPT-4 had been a genuine step-change, and GPT-4o followed with faster responses and more natural interaction. Many users were expecting GPT-5 to feel like giant step forward.

Instead, GPT-5 felt more evolutionary than revolutionary. For many users, the first interactions did not immediately justify the hype. That gap between expectation and experience shaped much of the early reaction.

Positive Reactions

There were, of course, strong positives. Many reviewers noted that GPT-5 showed clear improvements in:

• Complex reasoning and multi-step problem solving
• Long-context comprehension and retention
• Technical tasks such as coding, analysis, and structured planning

In more demanding scenarios, GPT-5 often performed better than its predecessors. For users working with large inputs or intricate constraints, the gains were real and measurable. Over longer sessions, GPT-5 tended to stay on track more reliably and required fewer corrections.

Negative Reactions and Frustrations

The criticism around GPT-5 was not about it being weak, but about it not feeling like the dramatic leap many had imagined.

One of the loudest reactions centred on personality. Many users bemoaned the lack of warmth and character in GPT-5 when compared to GPT-4o, which had become particularly well liked for its more natural, human-feeling responses. I noticed this myself. In side-by-side use, GPT-5’s responses often felt slightly off, more mechanical, and lacking what I can only describe as the “soul” that made GPT-4o so pleasant to interact with.

Another common complaint was inconsistency. GPT-5 introduced more automatic routing between faster and more deliberate modes of thinking. While this made sense from a system design perspective, it sometimes led to unpredictable behaviour. Users would see noticeably different quality from one interaction to the next, even with similar prompts.

Control was another major point of frustration. OpenAI removed the ability for users to explicitly choose between different models. While this simplifies things on paper, many users missed the ability to choose different models. Some users felt so passionate about the changes that they went as far as cancelling their subscriptions, not because GPT-5 was bad, but because they felt that the choice was being taken away from them by locking them into a single model. This backlash was significant enough that OpenAI listened and quickly reinstated access to GPT-4 and GPT-4o for users after the initial GPT-5 rollout, recognising how important those options were to their workflows and preferences

There was also a sense that the launch messaging outpaced the lived experience. Marketing language suggested a clear, unmistakable upgrade, but for many everyday tasks such as writing, summarisation, or casual exploration, GPT-5 did not feel dramatically better. In some cases, users felt it was harder to get the tone or precision they were used to.

The Perspective Problem

A recurring theme in the reaction to GPT-5 was perspective. Technically, GPT-5 is a very strong model. Viewed in isolation, it would be remarkable. Viewed in the context of what came before, it had the impossible task of exceeding not just prior models, but the emotional memory of discovering them.

As users, our expectations of OpenAI have grown alongside the models themselves. Each release trains us to expect a clear, immediate “wow” moment. When that moment does not arrive, even genuine progress can feel underwhelming.

Where GPT-5 Fits

Over time, many users began to appreciate GPT-5 for what it is rather than what they hoped it would be. In complex workflows, especially those involving reasoning, planning, or large volumes of information, GPT-5 shows its strengths. It is not always the most pleasant model to chat with, but it is often the most capable when the task demands depth.

This makes GPT-5 less of a universal upgrade and more of a specialised tool. It excels when you need it to think carefully, but it is not always the best choice for lightweight or creative interactions.

Final Thoughts

GPT-5’s reception says as much about us as it does about the model itself. OpenAI has set expectations incredibly high through years of rapid progress. Each new release is judged not just on its merits, but against a growing sense of what should be possible.

GPT-5 is not a disappointment, but it is a reminder. Progress at this level is incremental, not magical. As users, our expectations are huge, and sometimes unrealistically so. GPT-5 may not deliver an instant sense of wonder, but in the right contexts, it proves itself to be a powerful and serious piece of technology.