OpenAI Launches ChatGPT-5: Major AI Leap, Free to All Users
OpenAI has unveiled ChatGPT-5, its latest iteration of the flagship artificial intelligence model, describing it as a significant leap forward in AI capabilities. Released on Thursday, August 7, 2025, the new model boasts enhanced reasoning, task automation, and coding prowess, and is being rolled out free to all users, underscoring the intensified global competition in the rapidly evolving AI landscape.
This new generation of ChatGPT, which currently serves nearly 700 million users weekly, marks a substantial advancement. Sam Altman, OpenAI’s co-founder and chief executive, characterized ChatGPT-5 as “clearly a model that is generally intelligent,” representing a “significant step toward models that are really capable.” While acknowledging that true artificial general intelligence (AGI)—AI that can think and continuously learn like humans—remains a future goal, Altman emphasized the “huge improvement” in the current model’s capabilities.
Michelle Pokrass, a member of the development team, highlighted ChatGPT-5’s particular strength in acting as an “agent,” independently managing complex computer tasks. Altman further illustrated the progression, likening previous versions to educational stages: “GPT-3 felt to me like talking to a high school student… GPT-4 felt like you’re talking to a college student; GPT five is the first time that it really feels like talking to a PhD-level expert in any topic.”
A defining feature of this new era is expected to be “vibe-coding”—the ability to create software programs on demand. OpenAI executives demonstrated this by having the bot generate an application for learning the French language. Altman asserted that ChatGPT-5 leads the pack across various domains, including coding, writing, and healthcare, amidst fierce global competition from tech giants like Google and Microsoft, both of whom are investing billions into AI development. Despite the current strides, Altman anticipates “orders of magnitude more gains” on the path to AGI, a journey that will necessitate “eye-watering” investments in computational power.
Beyond raw capability, OpenAI has also prioritized safety and trustworthiness in ChatGPT-5. Alex Beutel, OpenAI’s safety research lead, explained that the model has been rigorously trained to be honest and provide “safe completions,” ensuring it delivers helpful answers without facilitating harmful missions. Evaluations were specifically designed to measure and mitigate the prevalence of deception.
The debut of ChatGPT-5 follows a period of strategic moves by OpenAI. Just prior to this release, the company announced a partnership with the US General Services Administration, allowing the US government to access a business-oriented version of ChatGPT Enterprise for a symbolic $1 per year. Furthermore, OpenAI recently released two new “open-weight language models,” gpt-oss-120b and gpt-oss-20b, which can be downloaded and modified by users. This move, challenging similar offerings from US and Chinese competitors, also addresses increasing pressure on the company to share the internal workings of its software, aligning with its original non-profit ethos.