Artificial Intelligence's New Advancement: A Comparative Look at GPT-5 and Its Rivals in the Market
OpenAI Unveils GPT-5: A New AI System with Improved Performance and Reduced Hallucinations
OpenAI has announced the release of GPT-5, a new AI system, designed to offer improved performance in coding and scientific knowledge benchmarks, while also reducing hallucinations in medical and factual queries.
GPT-5 is now available to all ChatGPT users, including those on the free tier. This AI system, which supports multimodal inputs with a large context window, includes numerous built-in tools such as web browsing, voice interaction, and integration with Gmail, Google Calendar, and Google Contacts for Pro users.
In the SWE-bench Verified coding benchmark, GPT-5 scored an impressive 74.9%, outperforming Claude Opus 4.1 at 74.5% and significantly outperforming Gemini 2.5 Pro at 59.6%. On the GPQA Diamond (PhD-level science) benchmark, GPT-5 Pro scored 89.4%, outperforming Claude Opus 4.1 at 80.9% and Grok 4 Heavy at 88.9%.
However, on some extended reasoning tasks, xAI’s Grok 4 Heavy slightly surpasses GPT-5. For instance, on a test spanning math, humanities, and natural sciences, Grok 4 Heavy scored 44.4%, while GPT-5 Pro scored 42% with tools.
GPT-5 incorporates enhanced safety measures, reducing hallucination rates, and improving adherence to instructions. According to OpenAI, the hallucination rate in GPT-5's medical benchmark is more than seven times lower than GPT-4o's.
New developer features include free-form function calling, verbosity controls for response detail, and a "reasoning effort" toggle to switch between fast answers and deeper analysis. GPT-5 Mini and GPT-5 Nano are also available for developers, each offering different latency and cost trade-offs.
The context window has expanded to 256,000 tokens in GPT-5, up from 200,000 in the o3 model. GPT-5 arrives with a unified experience for users, combining a real-time router that decides which approach to use based on conversation type, complexity, tool needs, and user intent.
Pricing for GPT-5 is $1.25 per million input tokens with a 90% cache discount and $10 per million output tokens. GPT-5 Mini is priced at $0.25 for input and $2 for output per million tokens, while GPT-5 Nano is $0.05 and $0.40, respectively.
The rollout begins immediately for all user tiers, with enterprise and education customers gaining access next week. Pro customers can use GPT-5 without limits. Plus, users get significantly more usage than free accounts, and free users switch to GPT-5 Mini after hitting their GPT-5 limit.
In addition to these features, GPT-5 offers a unified experience for users, combining a real-time router that decides which approach to use based on conversation type, complexity, tool needs, and user intent. It also introduces a new approach to sensitive requests called "safe completions."
With its improved performance in coding and scientific knowledge benchmarks, reduced hallucinations in medical and factual queries, and new features for developers and users, GPT-5 is set to redefine AI's role in our daily lives.
The announcement of GPT-5, a new AI system from OpenAI, brings advancements in technology, particularly in the areas of coding and scientific knowledge, with its impressive scores in SWE-bench Verified coding benchmark and the GPQA Diamond (PhD-level science) benchmark. Additionally, GPT-5 showcases the integration of artificial-intelligence with various gadgets such as web browsing, voice interaction, and integration with popular apps like Gmail, Google Calendar, and Google Contacts.
