OpenAI launches GPT-5, and it’s available to all ChatGPT users

In a major milestone for artificial intelligence, OpenAI has officially launched GPT-5, its most powerful and ambitious AI model to date. Unveiled on Thursday, GPT-5 is being hailed as the company’s first “unified” model — combining the reasoning depth of its o-series with the speed and fluency that made its GPT series so popular. The result is a next-generation engine powering ChatGPT that feels less like a chatbot, and more like an agent capable of real-world tasks.

“Having something like GPT-5 would be pretty much unimaginable at any previous time in history,” said OpenAI CEO Sam Altman in a press briefing. “This is the best model in the world.”

From Answers to Actions

While GPT-4 revolutionized chatbots with its ability to generate intelligent responses, GPT-5 takes a leap toward autonomy. The model allows ChatGPT to perform complex tasks on behalf of users — from coding entire software applications, to parsing medical test results, generating research briefs, or managing personal calendars.

Unlike previous versions that required users to fine-tune settings, GPT-5 comes with a real-time routing system that dynamically adjusts how it answers questions — either responding quickly or taking more time to think. This not only simplifies the user experience but also improves performance across a wider range of tasks.

Free Access for All

In a significant shift, GPT-5 is now the default model for all free users of ChatGPT, marking the first time OpenAI has given free-tier access to its flagship reasoning model.

“This is just one of the ways that I’m excited to live the mission,” said Nick Turley, VP of ChatGPT, referencing OpenAI’s goal to make advanced AI widely accessible.

Benchmarking the Frontier

GPT-5 arrives amid high expectations. Since its viral debut in 2022, ChatGPT has grown to over 700 million weekly users, making it one of the most widely adopted AI products on the planet. Its latest release is seen as a bellwether for where AI is headed — with the potential to impact industries from finance to healthcare.

Benchmark results show GPT-5 is leading or competitive in several categories:

• Coding: GPT-5 scored 74.9% on SWE-bench Verified, outperforming Claude Opus 4.1 (74.5%) and Gemini 2.5 Pro (59.6%).
• Science Reasoning: On GPQA Diamond, GPT-5 Pro achieved 89.4%, edging out Grok 4 Heavy (88.9%) and Claude Opus 4.1 (80.9%).
• Medical Accuracy: On HealthBench, GPT-5 (with reasoning) hallucinated only 1.6% of the time — far lower than GPT-4o (12.9%) or o3 (15.8%).

However, the model underperforms in some areas. For example, in navigating airline websites (Tau-bench), GPT-5 scored 63.5%, slightly behind o3 (64.8%). And on Humanity’s Last Exam, GPT-5 Pro scored 42%, just below Grok 4 Heavy’s 44.4%.

Creativity, Safety, and “Better Vibes”

According to OpenAI, GPT-5 also excels in creative writing, design, and subjective tasks. Turley noted that the model exhibits “better taste” and more natural expression in creative work.

Safety has also been enhanced. GPT-5 demonstrates a reduced tendency toward deception, and its hallucination rate in ChatGPT responses is down to 4.8%, compared to over 20% in previous models. It's also better at identifying and refusing unsafe requests, while remaining accessible to legitimate ones.

“The vibes of this model are really good,” said Turley.

New Tools for Users and Developers

ChatGPT users can now personalize interactions by choosing from four new model “personas”: Cynic, Robot, Listener, and Nerd, each offering a distinct conversational style.

For developers, GPT-5 will be available via API in three tiers:

• gpt-5
• gpt-5-mini
• gpt-5-nano

Pricing starts at $1.25 per million input tokens and $10 per million output tokens. Pro subscribers ($200/month) will get access to GPT-5 Pro, an enhanced version with more compute and extended reasoning capabilities.

Looking Ahead: GPT-5 and the Road to AGI

Altman has framed GPT-5 as a critical step toward artificial general intelligence (AGI) — a system capable of outperforming humans at most economically valuable work. Though not quite there yet, GPT-5’s ability to integrate reasoning, task execution, and natural dialogue suggests that AI agents are no longer a futuristic concept, but an emerging reality.

The release also comes just days after OpenAI launched gpt-oss, an open-weight reasoning model available for free — signaling a broader push toward openness and developer empowerment.

Still, GPT-5 is not a runaway lead in all categories. Benchmarks show it performs on par with other frontier models like Claude, Gemini, and Grok in certain areas. As always, real-world use cases will determine whether GPT-5 represents a true paradigm shift — or simply the next step in a longer AI journey.