OpenAI has officially announced its latest artificial intelligence model, GPT-5, which will be available to all users. It is now faster to respond, writes better code even for short prompts, and «hallucinates» less.
OpenAI has officially announced its latest artificial intelligence model, GPT-5, which will be available to all users. It is now faster to respond, writes better code even for short prompts, and «hallucinates» less.
«I tried going back to GPT-4, and it was pretty terrible,» OpenAI CEO Sam Altman said at the new model’s launch, comparing the jump from GPT-4 to GPT-5 to the iPhone’s transition from Pixel to Retina display.
«GPT-5 is the first time you really feel like you’re talking to an expert on any topic, like a PhD-level expert,» Altman said.
As Business Insider reports, the model will be available for free to all users. They will no longer have to choose a model themselves — GPT-5 will switch automatically, depending on the type of query and its complexity. GPT-5 will be available in standard, mini, and nano versions. Paid users will have higher usage limits.
OpenAI says the $200 per month Pro tier offers unlimited access to GPT-5, as well as GPT-5-pro, a more powerful version of the model, and GPT-5-thinking, which allows the model to process queries longer than usual. Pro users will still have access to the older models.
According to the developer blog, GPT-5 will cost developers using the API $1.25 per 1 million input tokens and $10 per 1 million output tokens.
Starting next week, Pro users will be able to connect their Gmail, Google Contacts, and Google Calendar to ChatGPT — with other tiers getting access later. «ChatGPT automatically knows when it’s most appropriate to link to them, so you don’t have to select them before you chat,» the company said in an email.
According to Sam Altman, GPT-5 is the company’s fastest model to date and has improved vibecoding capabilities. OpenAI claims that the model can «create beautiful and responsive websites, apps, and games with aesthetic considerations in just one command.»
GPT-5 produces better code, generates front-end UIs with minimal prompting, and demonstrates improvements in customizability, manageability, and complex query execution. GPT-5 also introduces «minimal» reasoning and an «extensibility» parameter to the API.
The company’s API will offer users all three models, along with additional controls to switch between detailed or direct answers. GPT-5 can store more information than previous models — the model has a context window of 256,000 tokens, an improvement over the 200,000-token context window available in the company’s previous o3 model. This means it can better understand long conversations, documents, or code without losing context.
Among other improvements, GPT provides more accurate and reliable answers to health-related questions, and is generally much less likely to «hallucinate,» that is, to make up facts.
After testing the GPT-5 model without access to the web, the researchers found that its level of hallucinations was 26% lower than that of the GPT-4o model. The level of hallucinations during reasoning in the GPT-5 model was 65% lower than that of the o3 model.
OpenAI says GPT-5 is the new standard in ChatGPT, replacing GPT-4o, OpenAI o3, OpenAI o4-mini, GPT-4.1, and GPT-4.5 for registered users. The model is rolling out today to all Plus, Pro, Team, and Free users, with Enterprise and Edu access coming in a week.
For users of the free version of ChatGPT, it may take a few days for the reasoning features to fully deploy. Once they reach their GPT-5 usage limit, they will be upgraded to GPT-5 mini, a smaller, faster, and higher-performance model.
