OpenAI has launched GPT-5, a new flagship AI model that will power the next generation of ChatGPT. Released on Thursday, GPT-5 is OpenAI’s first “unified” AI model, combining the reasoning abilities of its o-series models with the fast responses of its GPT series. This next-generation model marks a new era for ChatGPT and OpenAI, signaling the company’s broader ambitions to develop AI systems that function more like agents than chatbots.
While GPT-4 enabled AI chatbots to provide intelligent responses on a wide range of topics, GPT-5 allows ChatGPT to complete tasks on behalf of users, such as generating software applications, managing calendars, or creating research briefs. OpenAI has also made ChatGPT simpler to use with GPT-5. Instead of requiring users to adjust settings manually, GPT-5 includes a real-time router that determines the best way to respond—whether that means answering quickly or taking extra time to think through complex queries.
During a briefing with reporters, OpenAI CEO Sam Altman called GPT-5 “the best model in the world” and described it as a significant step toward artificial general intelligence (AGI), where AI can outperform humans in most economically valuable work. “Having something like GPT-5 would be pretty much unimaginable at any previous time in history,” Altman said.
Starting Thursday, GPT-5 is available to all free ChatGPT users as their default model. OpenAI’s VP of ChatGPT, Nick Turley, explained that this move aligns with the company’s mission to provide advanced AI to as many people as possible. Previously, more advanced models were restricted to paying subscribers.
GPT-5 is one of OpenAI’s most anticipated releases since ChatGPT debuted in 2022. Since then, ChatGPT has grown into one of the world’s most popular consumer products, reaching over 700 million users weekly—nearly 10% of the global population. Many view GPT-5 as a benchmark for AI progress, and its reception could influence Big Tech, Wall Street, and policymakers. Stakeholders are watching to see if GPT-5 delivers a leap in capabilities similar to GPT-4, which redefined expectations for AI.
OpenAI claims GPT-5 leads in several domains, slightly outperforming top models from Anthropic, Google DeepMind, and xAI on key benchmarks. However, it falls short in other areas. GPT-5 excels in coding, with the ability to generate entire software applications—a capability known as “vibe coding.” On SWE-bench Verified, a real-world coding test, GPT-5 scored 74.9%, just ahead of Anthropic’s Claude Opus 4.1 (74.5%) and Google DeepMind’s Gemini 2.5 Pro (59.6%).
On Humanity’s Last Exam, a challenging test covering math, humanities, and sciences, GPT-5 Pro (with extended reasoning) scored 42% with tools, slightly below xAI’s Grok 4 Heavy (44.4%). However, on GPQA Diamond, a PhD-level science test, GPT-5 Pro outperformed competitors with an 89.4% score, compared to Claude Opus 4.1 (80.9%) and Grok 4 Heavy (88.9%).
GPT-5 also improves in healthcare-related responses. On HealthBench Hard Hallucinations, GPT-5 hallucinated just 1.6% of the time, a significant improvement over GPT-4o (12.9%) and o3 (15.8%). While AI chatbots are not medical professionals, millions use them for health advice. OpenAI says GPT-5 is more proactive in flagging potential health concerns and helping users interpret medical information.
In creative fields, GPT-5 responds more naturally and exhibits “better taste” than other models, according to Turley. It also hallucinates less, with incorrect responses occurring only 4.8% of the time, compared to 22% for o3 and 20.6% for GPT-4o.
On Tau-bench, a test measuring AI’s ability to complete simulated online tasks, GPT-5 showed mixed results. It scored 63.5% on airline website navigation (below o3’s 64.8%) but 81.1% on retail website navigation (below Claude Opus 4.1’s 82.4%).
OpenAI emphasizes that GPT-5 is safer than previous models, exhibiting fewer deceptive behaviors and better distinguishing between harmful and harmless requests. Alex Beutel, OpenAI’s safety research lead, said this makes GPT-5 more transparent and trustworthy.
For consumers, ChatGPT now offers four new personality settings: Cynic, Robot, Listener, and Nerd. These adjust responses without requiring explicit user instructions. ChatGPT Plus subscribers ($20/month) get higher GPT-5 usage limits, while Pro subscribers ($200/month) have unlimited access to GPT-5 Pro, an enhanced version. Organizations on OpenAI’s Team, Edu, and Enterprise plans will gain default GPT-5 access next week.
Developers can access GPT-5 via OpenAI’s API in three sizes—gpt-5, gpt-5-mini, and gpt-5-nano—with varying reasoning capabilities. They can also control response verbosity. Pricing starts at $1.25 per million input tokens and $10 per million output tokens.
The launch follows OpenAI’s recent release of gpt-oss, an open-weight reasoning model that nearly matches the performance of o3 and o4-mini at a lower cost. While GPT-5 sets new standards in coding and other areas, it remains competitive with other frontier models in several benchmarks. Real-world applications will ultimately determine whether GPT-5 truly surpasses its rivals.