OpenAI's GPT-5 goes free: Enhanced coding, reasoning, and new features

OpenAI has begun rolling out its latest large language model, GPT-5, marking another significant milestone in the fiercely competitive race to develop the most advanced AI chatbot. The new model, which OpenAI asserts can deliver “PhD-level expertise” in complex domains like coding and writing, is now accessible to ChatGPT Plus, Pro, Team, and Free users, with Enterprise and Education users slated to receive access within the week. This bold claim echoes similar statements from figures like Elon Musk regarding his AI chatbot, Grok, integrated into X (formerly Twitter).

While some industry experts suggest GPT-5’s launch might prioritize maintaining market buzz over revolutionary leaps, its general-purpose capabilities include notably improved reasoning and robust coding skills. Whether OpenAI CEO Sam Altman’s assertions about GPT-5’s prowess hold true remains to be seen, but the model introduces several key features and functions worth examining.

GPT-5 is indeed free for the vast user base of ChatGPT, which currently serves approximately 700 million weekly users. However, access to the free version of ChatGPT is subject to usage caps. OpenAI’s business model relies on tiered subscriptions and enterprise integrations. For instance, the $20 per month ChatGPT Plus subscription offers substantially higher usage allowances. ChatGPT Pro subscribers, who pay $200 per month, gain unlimited access to GPT-5 and can utilize GPT-5 Pro, an even more advanced iteration of the model. Enterprise and Education users are promised “generous limits” designed to meet their organizational needs.

For the burgeoning community of programmers and developers who increasingly leverage AI for coding, GPT-5 aims to be an indispensable companion. OpenAI claims the new model significantly enhances coding proficiency. During demonstrations, company executives showcased GPT-5’s ability to generate custom, interactive applications—ranging from websites to games—on demand through simple prompts. While not flawless, GPT-5 represents a notable advancement in coding capabilities compared to its predecessors.

The new model also introduces four distinct preset “personalities” for user interaction: “cynic,” “robot,” “listener,” and “nerd.” These personalities, which can also be adjusted in ChatGPT settings, are designed to foster more natural and varied conversations, specifically addressing and reducing the tendency for excessive agreeableness observed in previous models like GPT-4o.

One of GPT-5’s most anticipated features is its deeper integration with external productivity applications. Building on existing connections with services like Google Drive, GPT-5 now integrates with Gmail and Google Calendar. This allows the AI chatbot to assist with a range of productivity tasks, including scheduling, reminders, and email follow-ups, by automatically pulling and referencing information from these connected apps. This feature, designed to function as a sophisticated personal AI agent, will initially roll out to Pro users.

A critical improvement in GPT-5, highlighted by OpenAI, is a significant reduction in factual errors, commonly known as “hallucinations.” These fabrications have long been an enduring challenge for AI systems, and while researchers acknowledge that completely eradicating hallucinations remains an elusive goal, the focus is on minimizing their frequency. According to OpenAI, GPT-5’s responses are approximately 45 percent less prone to factual inaccuracies compared to GPT-4o, though it can still hallucinate even during routine tasks.

Furthermore, GPT-5 shows enhanced capabilities in handling health-related queries, reflecting a growing trend of users seeking information from AI chatbots. OpenAI reports that GPT-5 scores higher than any previous model on the HealthBench test, acting more like an “active thought partner” by proactively flagging potential concerns and asking clarifying questions to provide more helpful answers. It is crucial to remember, however, that an AI chatbot is not a substitute for professional medical advice, and users should always consult a doctor for health-related decisions.

OpenAI emphasizes GPT-5’s improved ease of use, partly attributed to a “real router” feature. This intelligent system automatically determines the most appropriate underlying model to employ based on the complexity and context of a user’s prompt. Powered by a new “reasoning model,” GPT-5 engages in deeper cognitive processing to solve problems. Despite these advancements, the consensus remains that GPT-5 is more an evolution than a revolution, akin to incremental upgrades seen in new smartphone releases. While undeniably more intelligent, OpenAI’s overarching objective of achieving artificial general intelligence (AGI) still appears to be a distant horizon.