GPT-5 has been a positive experience for some developers. Jenny Wang, an investor and engineer who created the Alta personal styling agent, told WIRED that the GPT-5 model seems to do more complex tasks with one pass than any other model. She said it was similar to OpenAI’s o3 or 4o which she frequently uses for simple fixes and code generation. “like formatting, or if I want to create an API endpoint similar to what I already have,” Wang says.
Wang said that in her GPT-5 tests, she had asked the model for code to create a page of press for the website of her company, with specific design elements which would fit the aesthetic of the entire site. GPT-5 finished the job in one go, where in the previous, Wang’s prompts would have needed to be revised. However, there was one major error: “It hallucinated the URLs,” Wang says.
GPT-5, according to another developer speaking on condition of anonymity, because his employer did not authorize him to speak with the media, excels in solving difficult technical issues.
Developer’s current project involves writing an analysis network program that requires code isolation. “I basically presented my project and some paths I was considering, and GPT-5 took it all in and gave back a few recommendations along with a realistic timeline,” This is what the developer says. “I’m impressed.”
OpenAI enterprise customers Cursor Windsurf and Notion have all publicly acknowledged GPT-5 as having superior coding and reasoning abilities. OpenAI included many remarks from these customers and partners in its own blog post Announcement of the new model). Also, consider shared on X This is what it means “fast, thorough, and handles complex work 15 percent better than other models we’ve tested.”
But, within days of GPT-5’s release some developers started to complain online. GPT-5 was criticized for its coding capabilities, which many said were behind the curve. This is despite GPT-5 being a high-end model by one of the most popular AI companies in world.
“OpenAI’s GPT-5 is very good, but it seems like something that would have been released a year ago,” Kieran Klans, a programmer who is building an AI assistant to help with email inboxes. “Its coding capabilities remind me of Sonnet 3.5,” He also mentions an Anthropic-style model, which was launched in June of 2024.
Amir Salihefendić, founder of the startup company Doist, said in a social media post Cursor GPT-5 has been found by the user. “pretty underwhelming” This is what you should know “it’s especially bad at coding.” The release of GPT-4 was a traumatic experience for him. “Llama 4 moment,” Meta’s AI-model, which disappointed many in the AI community.
On X, developer Mckay Wrigley wrote GPT-5 has been dubbed as a “phenomenal everyday chat model,” But when it comes down to coding “I will still be using Claude Code + Opus.”
Other developers describe GPT-5 as “exhaustive”—at times helpful, but often irritating in its long-windedness. Wang was happy with her frontend code project for GPT-5. However, she noticed that the model she used was not the same as the original. “more redundant. It clearly could have come up with a cleaner or shorter solution.” The verbosity level of GPT-5 is adjustable. Users can request that it be more chatty, or do less reasoning for cheaper prices or better performance.
