OpenAI has introduced GPT-5 — the most powerful model in the history of ChatGPT

OpenAI officially launched GPT-5 — a new version of ChatGPT that promises to be more useful for real-world tasks. The model received a powerful upgrade across all key areas — from programming and medicine to poetry and visual analysis.

The main feature of GPT-5 is "smart thinking": the system itself decides when to give a quick answer and when to enable extended reasoning for complex questions. Users can directly say "think carefully" in their prompt, and the AI will switch to maximum depth mode.

What has been improved:
— coding: GPT-5 can create full websites, games, and apps from a single prompt, taking design aesthetics into account. In tests, it successfully assembled projects ranging from pixel art to interactive mini-games
— writing: the model has learned to write more expressively, maintain complex forms, and assist with business texts without excessive "plasticity"
— health: on HealthBench tests, GPT-5 became OpenAI's most accurate AI consultant, offering not diagnoses but help in formulating the right questions for doctors.

GPT-5 surpassed the company's previous achievements with 94.6% on a math olympiad without tools, 88% in code editing across different languages, and 84.2% on multimodal tasks.

GPT-5 in "thinking mode" makes ~80% fewer factual errors than OpenAI o3, and is twice as unlikely to give overly agreeable or "sycophantic" responses.

GPT-5 is already available to all ChatGPT users, including the free tier. Plus subscribers get more requests, while Pro users get access to the GPT-5 Pro version with maximum analysis depth.

Source: openai.com