OpenAI has officially announced GPT-5, its latest frontier model to power ChatGPT. The announcement follows the release of two open weight models, one of which can run on the Mac, and changes to encourage healthier ChatGPT use this week.

What is GPT-5?

Here’s what OpenAI has to say about GPT-5:

Today we’re introducing GPT‑5⁠, OpenAI’s smartest, fastest, most useful model yet, and a major step towards placing intelligence at the center of every business. GPT‑5 unites and exceeds OpenAI’s prior breakthroughs in frontier intelligence, spanning 4o, OpenAI o-series reasoning, agents, and advanced math⁠(opens in a new window) capabilities. […]

GPT‑5 delivers leaps in accuracy, speed, reasoning, context recognition, structured thinking, and problem-solving. The true magic will happen when businesses start applying GPT‑5 to imagine new use cases.

As OpenAI has suggested this year, GPT-5 is a unified model that includes capabilities that have been developed across multiple models before today:

GPT‑5 is a unified system with a smart, efficient model that answers most questions, a deeper reasoning model (GPT‑5 thinking) for harder problems, and a real‑time router that quickly decides which to use based on conversation type, complexity, tool needs, and your explicit intent (for example, if you say “think hard about this” in the prompt). The router is continuously trained on real signals, including when users switch models, preference rates for responses, and measured correctness, improving over time. Once usage limits are reached, a mini version of each model handles remaining queries. In the near future, we plan to integrate these capabilities into a single model.

OpenAI will deprecate its previous models in favor of its new, unified frontier model. GPT-5 is also a more performant and more accurate model, according to OpenAI:

In our evaluations, GPT‑5 (with thinking) performs better than OpenAI o3 with 50-80% less output tokens across capabilities, including visual reasoning, agentic coding, and graduate-level scientific problem solving. [..]

GPT‑5 is significantly less likely to hallucinate than our previous models.  With web search enabled on anonymized prompts representative of ChatGPT production traffic, GPT‑5’s responses are ~45% less likely to contain a factual error than GPT‑4o, and when thinking, GPT‑5’s responses are ~80% less likely to contain a factual error than OpenAI o3.

Notably, GPT-5 is now the default model used by ChatGPT:

GPT‑5 is the new default in ChatGPT, replacing GPT‑4o, OpenAI o3, OpenAI o4-mini, GPT‑4.1, and GPT‑4.5 for signed-in users. Just open ChatGPT and type your question; GPT‑5 handles the rest, applying reasoning automatically when the response would benefit from it. Paid users can still select “GPT‑5 Thinking” from the model picker, or type something like ‘think hard about this’ in the prompt to ensure reasoning is used when generating a response.

GPT-5 availability

OpenAI says GPT-5 is available starting today starting for ChatGPT Team, Pro, Plus and free tier users and through the OpenAI API. Paid customers will have access to more usage, while free users will GPT-5 mini after hitting usage limits.