OpenAI launched GPT-5 on Thursday, a unified model that merges the reasoning power of its o-series with the fast responses of its GPT series. Sam Altman called it "the best model in the world" during a briefing with reporters. ChatGPT will use GPT-5 as the default for all free users starting immediately. The company says it wants AI that behaves more like an agent than a simple chatbot, and this model is the latest step in that direction.
While GPT-4 answered questions on many topics, GPT-5 is designed to complete full tasks, such as building software applications, navigating a user's calendar, or creating research briefs. OpenAI built a real-time router that decides how to offer the best answer, whether that's responding quickly or taking additional time to "think" through complex problems. Users no longer need to manually select between speed and depth modes.
Benchmark Numbers Tell a Split Story
OpenAI claims GPT-5 is state-of-the-art in several domains, slightly edging out leading AI models from Anthropic, Google DeepMind, and Elon Musk's xAI on key benchmarks. However, GPT-5 slightly underperforms frontier AI models in other areas. The company says GPT-5 offers frontier-level performance around coding; Altman said the model specifically excels at spinning up entire software applications on demand, in what has become known as "vibe coding."
On SWE-bench Verified - a test of real-world coding tasks pulled from GitHub - GPT-5 scores 74.9 percent on its first attempt. That means GPT-5 just outperforms Anthropic's latest Claude Opus 4.1 model, which scored 74.5 percent, and Google DeepMind's Gemini 2.5 Pro, which scored 59.6 percent. The gap in coding performance represents a clear win for developers who rely on AI assistance to build software.
On Humanity's Last Exam - a difficult test measuring AI model performance across math, humanities, and the natural sciences - a version of GPT-5 with extended reasoning (GPT-5 Pro) scored 42 percent when using tools. That's slightly less than xAI was able to achieve with Grok 4 Heavy, which scored 44.4 percent on the test. The mixed results suggest that benchmark leadership depends on the specific task at hand.
On GPQA Diamond - a test of PhD-level science questions - GPT-5 Pro scored 89.4 percent on its first try, outperforming Claude Opus 4.1, which scored 80.9 percent, and Grok 4 Heavy, which scored 88.9 percent. OpenAI says GPT-5 is better for answering health-related questions. On a test measuring accuracy in AI model responses around healthcare topics, HealthBench Hard Hallucinations, OpenAI says GPT-5 (with thinking) hallucinates just 1.6% of the time. This is far lower than the company's previous GPT-4o and o3 models, which scored 12.9% and 15.8%, respectively.
While AI chatbots are not medical professionals, millions of people are using them for health advice. In response to this phenomenon, the company says GPT-5 is more proactive about flagging potential health concerns and helping users parse medical results. In responses to ChatGPT prompts, OpenAI found that GPT-5 (with thinking) hallucinates and responds with incorrect information 4.8% of the time. That's a reduction from o3 and GPT-4o, which score hallucination rates of 22% and 20.6%, respectively, on the test.
Pricing Makes GPT-5 Accessible to Developers
For developers, GPT-5 is coming to OpenAI's API in three sizes - gpt-5, gpt-5-mini, and gpt-5-nano - which will spend more or less time "reasoning" through tasks. Developers can also now control verbosity in the OpenAI API, deciding how long or short an AI model's responses should be. The base model of GPT-5 will cost developers $1.25 per million input tokens (roughly 750,000 words, longer than the entire "Lord of the Rings" series) and $10 per million output tokens.
That pricing represents a reduction compared to previous models, making large-scale deployments more affordable. Subscribers to ChatGPT's $20-per-month Plus plan get higher usage limits for GPT-5 than free users. Meanwhile, $200-per-month Pro subscribers will have unlimited access to GPT-5, as well as a souped-up version called GPT-5 Pro that uses additional computational resources to produce better answers. Organizations on OpenAI's Team, Edu, and Enterprise plans will gain access to GPT-5 as their default model next week. The tiered access model reflects OpenAI's strategy of balancing accessibility with revenue generation.
GPT-5 also makes its debut alongside gpt-oss, an open-weight reasoning model that developers and enterprises can download for free and run at a fraction of the cost. The open model nearly matched the abilities of OpenAI's previous top models, o3 and o4-mini, but GPT-5 sets a new standard for frontier performance in some areas, such as coding. This combination of proprietary power and open alternatives gives developers flexibility in choosing the right tool for their needs.
However, GPT-5 seems to be roughly on par with other frontier AI models in several areas. Benchmarks, of course, only tell part of the story for any AI model, and it remains to be seen how developers will use GPT-5 in the real world, and whether the model is truly a step above the competition. The gap between lab scores and everyday reliability is wide, and early customer feedback will matter more than benchmark tables.
Safety, Regulation, and Competition
The model's safety features represent another point of contention. OpenAI claims GPT-5 is less deceptive than previous models, but early testers have reported unexpected behaviors in niche use cases. Alex Beutel, OpenAI's safety research lead, said reducing deception improves not only the safety of GPT-5, but also the user experience, creating a model that is more "transparent and honest in ways users can trust." Beutel also notes GPT-5 is better at discerning between bad actors who are trying to misuse ChatGPT and users making harmless requests. This results in GPT-5 being able to refuse more unsafe questions, while offering fewer rejections to users seeking harmless information.
Regulators are catching up. European lawmakers are finalizing rules that could limit how AI models like GPT-5 are deployed in consumer products. The U.S. Congress is considering similar legislation. OpenAI's speed with a widely available model may shape compliance norms globally. Competitors such as Anthropic and xAI are already preparing responses of their own, and the pressure to match or exceed GPT-5's capabilities is only intensifying.
According to TechCrunch's detailed analysis, the model's reception by early adopters will determine whether GPT-5 becomes a lasting platform upgrade or a temporary novelty. OpenAI's distribution strategy, which puts GPT-5 in front of 700 million weekly ChatGPT users, suggests the company is prioritizing market share over immediate revenue maximization. Whether that bet pays off depends on how real-world performance aligns with the benchmark numbers. For more AI updates, see our AI desk and the benchmark tracker. The race continues, and the tests are just beginning.