ChatGPT-5 Underwhelming You? Here’s What It Can Do That Older Models Couldn’t — And Where Other AI Chatbots Still Shine
Last Thursday, OpenAI unveiled its latest version of the highly popular AI chatbot, ChatGPT. At the unveiling, Sam Altman, CEO of OpenAI, made some remarkable claims about GPT-5.
“GPT-3 was sort of like talking to a high school student: There were flashes of brilliance, lots of annoyance, but people started to use it and get some value out of it,” he said. “With GPT-4, maybe it was like talking to a college student: real intelligence, real utility. But with GPT-5, now it’s like talking to an expert: a legitimate Ph.D.-level expert in anything, any area you need, on demand that can help you with whatever your goals are.”
While OpenAI insists that GPT-5 is its most advanced AI yet, showcasing more expertise and capability, and fewer hallucinations and mistakes compared to previous models, not all users are convinced. Many were upset when OpenAI initially hid its other models, including GPT-4o, upon releasing GPT-5—though this decision has since been reversed. Additionally, some users have critiqued GPT-5’s conversational style, describing it as “cold” or “blunt” compared to its predecessors, and likened its responses to those of an “overworked secretary.” Others have highlighted the model’s stricter safety and refusal policies despite transparency improvements, noting that competitors like Anthropic’s Claude, Google’s Gemini, and Meta’s Llama still excel in certain tasks like live coding or customizable deployments.
If you find yourself underwhelmed by the arrival of GPT-5, don’t worry. We’ve highlighted several advantages GPT-5 holds over other AI models and pointed out where other models continue to shine.
1. Unified, Adaptive Reasoning
In its GPT-5 unveiling, OpenAI highlighted that previous versions of ChatGPT required users to select among different models for various tasks. For instance, users might have chosen “o3” for tasks demanding more research, while “4o” could deliver smart answers quickly. The main innovation with GPT-5 is its claim to eliminate the need to choose between these models. OpenAI suggests that GPT-5 is an all-around option, delivering both rapid responses and in-depth analysis, with particular strengths in coding and mathematics.
2. More Accurate Answers with Fewer Hallucinations
OpenAI states that GPT-5 excels in extended reasoning, with significantly fewer hallucinations and improved accuracy. For example, HealthBench scores reveal up to 80% fewer factual errors in complex scenarios compared to earlier versions, especially compared to GPT-o3, which was known for its precision despite slower response times.
3. Rich Personalization
GPT-5 introduces customizable personalities (Cynic, Robot, Listener, Nerd) and enhanced voice features, enabling users to adjust the chatbot’s tone and style without needing complex prompt engineering. The user interface now allows for adaptive color schemes and conversation settings. OpenAI asserts that the voice model sounds “incredibly natural” and can dynamically adjust its speech speed, proving ideal for language learning. Moreover, memory enhancements enable ChatGPT to learn from users over time and grasp their goals. Soon, paid users will enjoy access to Gmail and Google Calendar, allowing ChatGPT to plan schedules, integrate appointments, and manage emails, making it far more “useful and personal.”
4. Workflow Integration
GPT-5 supports direct integration with applications like Gmail and Google Calendar, letting the chatbot manage schedules and emails as instructed. Businesses can tailor GPT-5 with proprietary data, transforming it into a personalized digital assistant rather than a generic chatbot. In delivering actual work, OpenAI describes GPT-5 as “the best coding model on the market today,” capable of writing full computer programs independently. It can operate autonomously for prolonged periods, with a “self-improvement loop” allowing it to refine projects by sending detected errors back to itself for code iteration.
Despite its robustness and flexibility, GPT-5 is not the sole player in the field:
- Claude Opus 4.1 from Anthropic is well-regarded for autonomous coding and agentic workflows. Notable features like “Claude Code” include integrated code review and real-time security, providing specialized developer tools absent in GPT-5. Claude offers live code visualization within its integrated development environment (IDE), enhancing clarity for engineering teams. It also maintains artifacts locally to track long-running tasks across sessions, potentially surpassing GPT-5’s current workflow persistence in some circumstances.
- Google’s Gemini 2.5 Pro excels in seamless video analysis, which remains an evolving area for GPT-5. Gemini’s multimodal engine efficiently handles short videos, diagrams, and complex visual reasoning, offering profound contextual understanding. Its integration with Google’s Search offers real-time internet data access, whereas GPT-5 may struggle with real-time data or minute-by-minute market changes.
- Llama 3 from Meta caters to research settings and organizations with stringent privacy requirements, offering complete open-source customization. Users maintain full control over security and deployment, allowing for unique customization. Llama 3 performs excellently in managing lengthy documents and multi-turn dialogues, providing flexibility for internal infrastructure operations and expediting experimentation and enhancement.
Despite the enhancements OpenAI advocates for ChatGPT-5, the AI arms race remains very much in full swing. Anthropic, Google, and Meta continue to compete strongly in areas such as customization, coding, and live updates. As always, it is prudent to experiment with different models to determine the optimal model and workflow that best accommodates your needs.
