New Version of GPT-5 Shockingly Lags Behind in Innovation, Here's the Lowdown
OpenAI's latest offering, GPT-5, has created a stir in the AI community with its live-streamed unveiling on Thursday, 10 a.m. Pacific Time. This advanced language model is set to revolutionise the landscape of AI, offering significant improvements over previous GPT versions and competitive models.
Key features of GPT-5 include smart model routing, reduced hallucinations, improved honesty, multi-task proficiency, an extremely large context window, enhanced customizability, and enterprise adoption. The model uses a real-time router to automatically select the best sub-model for a given task, reducing the need for user intervention. It also has a significantly lower rate of misleading or overconfident outputs compared to earlier OpenAI models, improving user trust.
GPT-5's performance on benchmarks in coding, advanced math, visual reasoning, and real-world agentic tool use is state-of-the-art. It can handle long chains of tool calls both sequentially and in parallel, making it a versatile tool for a wide range of tasks.
In comparison to other major AI models like Google Gemini, Anthropic's Claude, and Elon Musk's xAI Grok, GPT-5 appears to surpass or at least match them in depth of reasoning, versatility, and latency. While Claude focuses on safe and ethical AI use, GPT-5 reduces hallucinations even more effectively and emphasizes honest communication about its limits. Grok, although emerging, is less mature in capability and deployment scale, particularly for complex, multi-agent workflows and professional tasks.
OpenAI will simplify its model selection process with the introduction of GPT-5, allowing the model to decide which model is best for a user's query. Users can access a free version of GPT-5, but with limits, and those paying at least $20 a month can use the full capacity. Unlimited use is reserved for Pro users paying $200 a month.
The presentation of GPT-5 emphasised its programming capabilities, with the goal of improving AI towards superintelligence. However, it was more evolutionary than revolutionary, as some competitors like Elon Musk's xAI Grok have already shown superior performance in certain areas.
OpenAI's CEO, Sam Altman, claimed that GPT-5 is a "significant step on the path to AGI" where AGI refers to a general super-intelligence that surpasses humans in all areas. Despite this, Altman admitted to feeling "useless" as a human after using GPT-5, highlighting the model's advanced capabilities.
The rollout of GPT-5 addresses a significant disadvantage for OpenAI in the competition for end-users, as Microsoft and Google can more easily integrate their AI models into existing applications like Outlook or Gmail. OpenAI has also announced that ChatGPT can be connected to a Gmail account and Google Calendar in the future, expanding its role as a personal assistant.
However, there was an embarrassing mistake or intentional deception in the chart OpenAI showed comparing GPT-5 to its predecessor models. The exact nature of this error is yet to be clarified.
In conclusion, GPT-5 stands out as highly capable, versatile, and developer-friendly—with state-of-the-art reasoning, honesty, and multi-task performance—leading the AI model field as of mid-2025 across diverse domains including coding, scientific problem-solving, enterprise automation, and multimodal understanding.
What is the significant improvement in GPT-5 over previous models regarding reducing the rate of misleading outputs?The technology behind GPT-5, artificial-intelligence, showcases its enhanced honesty in its responses, ensuring a more reliable user experience compared to earlier OpenAI models.
