San Francisco, Aug 10: OpenAI on Friday rolled out GPT-5, its latest large language model, making it free to all ChatGPT users for the first time, though free-tier access will be capped. The company is pitching the model as its most capable yet in coding and reasoning, with CEO Sam Altman calling it a “pretty significant step” toward artificial general intelligence (AGI).
The model is now available to ChatGPT Plus, Pro, Team and Free users, with Enterprise and Education customers to follow in a week. Subscribers to ChatGPT Plus, priced at $20 a month, will get higher usage limits, while $200-a-month Pro users will have unlimited access to GPT-5 and an advanced GPT-5 Pro version with extended reasoning. Developers can also access the model via OpenAI’s API, which comes in three sizes – gpt-5, gpt-5-mini, and gpt-5-nano – and in a non-reasoning “gpt-5-chat-latest” variant.
Marketed as OpenAI’s “best model yet for coding and agentic tasks,” GPT-5 offers enhanced accuracy, speed, math capabilities and a “real router” feature that selects the optimal model for a task. It introduces four preset chatbot personalities – “cynic,” “robot,” “listener” and “nerd” – aimed at making conversations more natural and reducing sycophantic responses. The model also integrates with Gmail and Google Calendar for scheduling and reminders, though this will initially be limited to Pro users.
OpenAI says GPT-5 reduces “hallucinations” – factual inaccuracies common in AI – by about 45% compared to GPT-4o, and ranks highest among its models for health-related queries, though the company cautions it is not a medical substitute.
Altman highlighted India’s rapid adoption of AI during the launch, describing it as OpenAI’s second-largest market after the United States. “It may well become our largest market,” he said. “It’s incredibly fast-growing, but what users are doing with AI, what citizens of India are doing with AI, is really quite remarkable.”
Some analysts see GPT-5 as more of an incremental upgrade than a breakthrough, noting that AGI remains a long-term goal. But Altman said the model represents the first time one of OpenAI’s mainline systems “has felt like you can ask a legitimate expert, like a PhD-level expert, anything.”
