Claude makes ChatGPT look like a robot with a fake mustache.

I had enough of ChatGPT congratulating mean, suggesting I am crazy or over dramatic and generally getting EVERYTHING some version of wrong. I’ve been using Claude for a month now and I couldn’t be happier. Never going back. Well done Anthrophic.

I currently maintain premium subscriptions to Claude, Cursor, and ChatGPT. In my experience, Claude Opus 4.7 is technically very impressive… but the question is whether it justifies the cost. Unfortunately, I don’t believe it does.For example, this evening I asked Claude to review a C# file I created (approximately 6,000 lines of code) to identify performance optimization opportunities. It provided some useful insights, so I then asked it to implement the suggested improvements. However, it only completed about half of the task before my token limit was reached… after roughly 15 minutes of use. At that point, I was required to wait approximately five hours for the limit to reset.By comparison, I can use ChatGPT (Codex) continuously throughout the day with no such interruptions, while achieving comparable results. To Claude’s credit, Opus did identify a few optimizations that Codex initially missed, but the marginal gains are outweighed by the severe usage limitations.When comparing Claude to alternatives such as Codex, Cursor, Gemini, and Grok, it becomes difficult to justify the premium subscription. If reliability and sustained productivity are important, other platforms… particularly Codex, offers a significantly better experience.Using Claude Sonnet instead of Opus does allow for more efficient token usage, but that raises another concern: why pay for access to Opus if it cannot support a full working session?I’m giving this product 3 stars based on the capabilities of Opus itself, but deducting 2 stars due to the pricing and token model. The current structure feels less like a subscription and more like a pay-per-result system. Running out of tokens mid-task and being prompted to pay more for additional usage is frustrating… especially when even a single code review and partial edit cannot be completed without extended waiting periods.In practical terms, completing a single task may require multiple sessions spaced hours apart, which is not a viable workflow. Competing platforms simply do not impose these limitations.If the goal is to see continued improvement across AI tools, it makes sense to support platforms that offer both strong performance and reasonable usability. In that regard, ChatGPT Codex stands out… it delivers excellent results with far fewer restrictions under the same subscription model.Do better Claude, stop being stingy with the tokens, absolutely will be the reason you lose a customer base as become become more familiar with AI and realize your competitors have a better, more sustainable model that can be used all day vs. Few minutes here and there, every 5 damn hours. :(

I am giving you my email and name. Why do you need my phone number as well? A person who wants to stay anonymous would use a temporary number anyway, so there’s no reason whatsoever to ask for the user’s phone unless you are in the data collecting and selling business.