OpenAI has launched the highly anticipated fifth generation of its flagship AI tool, ChatGPT, highlighting major advancements in artificial intelligence as the global competition in the tech space intensifies.

The new ChatGPT-5 is being made available to all users at no cost, with OpenAI revealing during a press briefing that the tool is currently used by nearly 700 million people each week.

OpenAI’s co-founder and CEO, Sam Altman, described this latest version as “clearly a model that is generally intelligent,” while noting that the company is still working toward achieving artificial general intelligence (AGI) — a form of AI that can understand and learn like a human.

“This model doesn’t learn in real-time from the data it encounters post-deployment, which is a capability I associate with true AGI,” Altman said. “Nonetheless, its abilities mark a substantial leap forward.”

Experts have long predicted a new era shaped by powerful AI tools capable of transforming daily work and entertainment. Meta’s CEO Mark Zuckerberg recently noted that the progress in AI is rapidly bringing the development of superintelligence closer to reality, calling it the start of “a new era for humanity.”

Altman emphasized that while there’s still a long road to AGI, OpenAI plans to continue investing heavily in computing power to drive future breakthroughs.

Tech giants including Amazon, Google, Meta, Microsoft, and Elon Musk’s xAI have poured significant resources into AI development since the original ChatGPT made headlines in late 2022. Earlier this year, Chinese firm DeepSeek also made waves with a model delivering impressive results using more affordable chips.

– ‘Like a PhD Expert’ –

Amid intense international competition, Altman said that ChatGPT-5 now leads the field in areas like software coding, content creation, and healthcare applications.

“GPT-3 felt like speaking to a high school student. GPT-4 was more like a college student. GPT-5, for the first time, feels like engaging with a PhD-level expert — on virtually any topic,” he explained.

He predicted that the ability to generate software from simple prompts — a concept referred to as “vibe-coding” — would define the capabilities of ChatGPT-5.

British AI expert Simon Willison, who had early access to the new model, shared his thoughts in a blog post: “My verdict: it’s just good at stuff. It doesn’t feel like a dramatic leap, but it exudes reliability — it rarely fails, and often impresses.”

However, Elon Musk expressed a different view, claiming that his Grok 4 Heavy AI model is “smarter” than ChatGPT-5.

– Building Trustworthy AI –

OpenAI said the latest model was designed to be reliable and avoid engaging in or assisting with harmful activities. According to Alex Beutel, who leads safety research at OpenAI, the model was trained to provide helpful and honest responses.

“We developed evaluations to measure potential deception and trained the model for honesty,” Beutel explained. He added that the system was engineered to generate only “safe completions,” avoiding details that could be misused.

In addition to ChatGPT-5, OpenAI also introduced two new downloadable AI models that users can modify — a strategic move aimed at competing with similar tools offered by other tech firms.

These “open-weight language models” reflect increasing calls for OpenAI to embrace transparency, in line with its origins as a nonprofit entity committed to openness in AI development.