OpenAI Officially Launches GPT-5: A Major Leap Toward AGI
OpenAI Officially Launches GPT-5: A Major Leap Toward AGI
OpenAI has officially rolled out GPT-5, the newest generation of its flagship language model, to both free and paid ChatGPT users. The long-anticipated update marks one of the company’s biggest steps toward artificial general intelligence (AGI) — though CEO Sam Altman says we’re not quite there yet.
During a press briefing, Altman called GPT-5 “a significant step along the path to AGI,” describing it as “clearly a model that is generally intelligent,” while acknowledging it still lacks core AGI traits like continuous learning after deployment. In OpenAI’s charter, AGI is defined as “a highly autonomous system that outperforms humans at most economically valuable work.”
Smarter, Faster, More Accurate
OpenAI says GPT-5 delivers higher accuracy, faster responses, and a lower hallucination rate than any previous model. Altman compared the leap from GPT-4 to GPT-5 to Apple’s shift from pixelated displays to Retina resolution:
“GPT-5 is the first time it really feels like talking to a PhD-level expert in any topic,” Altman noted.
The update introduces two new variants:
-
GPT-5-mini — Lightweight and faster for everyday use.
-
GPT-5-nano — An API-exclusive model that is cheaper and even quicker.
Free ChatGPT users will get GPT-5 and GPT-5-mini, while Plus subscribers enjoy higher usage limits. The $200/month Pro tier includes unlimited GPT-5 access, GPT-5-pro (a more powerful model), and GPT-5-thinking (capable of extended reasoning time). The chat interface now automatically chooses the best model for each query.
Developer Pricing & API Features
According to OpenAI’s launch blog, API pricing is as follows:
-
GPT-5: $1.25/1M input tokens, $10/1M output tokens
-
GPT-5-mini: $0.25/1M input tokens, $2/1M output tokens
-
GPT-5-nano: $0.05/1M input tokens, $0.40/1M output tokens
With a 256,000-token context window, GPT-5 can retain and process far more information than older models, making it ideal for long conversations, codebases, or multi-document analysis.
Performance Benchmarks
OpenAI reports that GPT-5 outperforms prior models in multiple coding tests:
-
SWE-Bench Verified: 74.9%
-
SWE-Lancer (GPT-5-thinking): 55%
-
Aider Polyglot: 88%
In live demos, GPT-5 quickly built a fully interactive, themed French-learning web app complete with quizzes, flashcards, and daily progress tracking.
New Features for ChatGPT Users
Starting next week, Pro users can link Gmail, Google Contacts, and Google Calendar directly to ChatGPT, with other tiers gaining access later. ChatGPT will automatically pull relevant info without users needing to prompt it.
Customization is also expanding: users can now choose chat colors and select from preset personalities — Cynic, Robot, Listener, and Nerd — which will integrate with Advanced Voice Mode.
Healthcare, Safety & Reduced Hallucinations
OpenAI claims GPT-5 is their best model yet for health-related questions, outperforming prior models in benchmarks like HealthBench and HealthBench Hard. Safety teams have reduced hallucination rates by 65% compared to the o3 model, and implemented “safe completions” to prevent harmful outputs. Over 5,000 hours of red-teaming were conducted to strengthen system reliability.
Massive Global Adoption
OpenAI reports 700 million weekly ChatGPT users, including 5 million paying businesses and 4 million developers using its API.
“The vibes of this model are really good,” said ChatGPT head Nick Turley. “Everyday users will feel the difference immediately.”
With GPT-5 now live, OpenAI has firmly reinforced its position at the forefront of the artificial intelligence revolution — and it’s one step closer to the AGI future Altman has long envisioned.
