Developers Say GPT-5 Is a Mixed Bag

Some developers say they have had a great deal of positive experiences with GPT-5 so far. Jenny Wang, an engineer, investor, and peelder of the personal stylingagent told that the model seems to be better in completing complex codening tasks in one school or other models. She compared with Openai's O3 and 4o, that they often used for code-generation and straight fixes “as formatting, or if I can create an API endpoint with what I already say,” saying cheek.

In their testing of GPT-5, Wang says they asked the model to generate code for the website of their business, including specific design elements that would match the aesthetic of the site. GPT-5 completed the task in one, while in the past, cheek would have to have their prompts in the process. However, there was one important mistake, although: “the hallucinated the URLs,” says Wang.

Another developer, who spoke on the condition of anonymity, as their employer has not authorized to speak to the press, GPT-5 excels says deep technical problems.

The current hobby project writes a program tool writes a programatic network analysis tools, which would require code isolation for safety goals. “I basically my project and some paths presented, and GPT-5 took it all and gave a few recommendations with a realistic timeline,” explains the developer. “I'm impressed.”

A handful of Openai's Enterprise Partners and customers, including cursor, Windsurf, and understand, have public vpt-5's encryption and reasoning skills. (Openai included a lot of these comments in their own Blog post announce the new model.) Notion too Shared on x So it's “Rap, thorough, and treats complex work 15 percent better than other models we have tested.”

But within days of the relitians of GPT-5 waves some developers in online with complaints. Many said that the Coding of GPT-5 should the curve seemed for what should be a state-of-the art, ultra-in state of the world's Buzziest AI business.

“Openai's GPT-5 is very good, but it looks like something that would be a year ago,” Says Kieran Klaassen, an AI Wizard for e-mailbuses. “The encoding opportunities remind me of Sonnet 3.5, he adds, reference to an anthropic model launched in June 2024.

AMIR SALIHEFENDIć, founder of the start-up business doist, said in a social media post That he has used in the cursor and found the “pretty underwill” “and that” it is mainly bad at encrypass. “He said that the release of GPT-4 felt like a” Llama 4-moment, “Referring to Meta's AI model, which also had disappointed in the AI ​​community.

ON X, Developer McKay Wrigley yell That GPT-5 is a “Fenomenal every day chat model,” but when it comes to coding, “I will still use the Claude code + opus.”

Other developers describe GPT-5 or “exhausting” -at times helpful, but often irritating in their long windens. Wang, who was generally with the Fronted Coding Project to GPT-5, says they were noticed the model. It could be with a cleaner or shorter solution. ” (KooPor point out that the VPT-5 Verbosity can be adjusted so users can ask it to be less conversation or even less reasoning in exchange for better performance or cheaper prices.)