OpenAI unveiled ChatGPT 5, its most advanced AI model yet, on August 7, 2025, making it available to all users, including those on the free plan, as announced by CNBC. The rollout, which began Thursday, introduces a reasoning-focused model that “thinks” through problems before responding, a first for free-tier users. “It’s like having a team of Ph.D.-level experts in your pocket,” OpenAI CEO Sam Altman said during a media briefing, comparing the leap from GPT-4 to GPT-5 to switching from a pixelated phone screen to a retina display, per NBC News.
ChatGPT 5 replaces older models like GPT-4o and o3, streamlining the interface into a single, auto-switching system that adjusts based on query complexity, per OpenAI’s blog. It boasts a 256,000-token context window, allowing it to process longer conversations or documents, such as expense reports or codebases, without losing context, according to Wired. The model is less prone to “hallucinations,” with 45% fewer factual errors than GPT-4o when web searching, and up to 80% fewer when using its “thinking” mode, per Ars Technica. Free users get up to 10 messages every 5 hours, while Plus users ($20/month) can send 80 messages every 3 hours, and Pro users ($200/month) enjoy unlimited access, per OpenAI’s Help Center.
New features enhance usability across writing, coding, and health queries. “GPT-5 is our most capable writing collaborator yet,” OpenAI noted, citing its ability to draft compelling emails or sustain unrhymed iambic pentameter, per OpenAI’s website. For coders, it generates deployable websites with minimal prompts, scoring high on the SWE-Bench Verified benchmark, per The Washington Post. Pro users will soon connect Gmail, Google Contacts, and Calendar, with ChatGPT referencing them automatically, per Wired. Users can also pick from four personalities—Cynic, Robot, Listener, or Nerd—and customize chat colors, per Ars Technica.
Key highlights of ChatGPT 5 include:
- Advanced Reasoning: Breaks down complex tasks like logic puzzles or financial analysis.
- Expanded Memory: Handles up to 256,000 tokens for longer, coherent conversations.
- Free Access: Available to all, with higher limits for Plus and Pro subscribers.
Despite its strengths, limitations persist. A live demo showed GPT-5 incorrectly explaining the Bernoulli effect on airplane wings, and OpenAI acknowledged “key limitations in areas like persistent memory,” per NBC News. Safety remains a priority, with 5,000 hours of testing and a new “safe completions” framework to handle sensitive queries, per CNBC. The model is rolling out globally across web, iOS, and Android, with Enterprise and Edu access starting in a week, per OpenAI. Users can try it now by visiting chat.openai.com.
