OpenAI releases the next generation AI model GPT-5

OpenAI launched its latest artificial intelligence model, GPT-5, on August 7. This is the company's first "integrated" AI system, combining the reasoning capabilities of the o series models with the fast response capabilities of the GPT series.

OpenAI CEO Sam Altman referred to GPT-5 as "the best model in the world", representing a "significant upgrade" compared to previous models. He stated that this marks an important step for the company on the path to achieving general artificial intelligence (AGI).

GPT-5 has performed exceptionally well in several benchmark tests. In the SWE-bench Verified code test, its accuracy reached 74.9%, slightly higher than Anthropic's latest Claude Opus 4.1 model. Meanwhile, GPT-5's misinformation rate is only 4.8%, far lower than the 20.6% of the previous model, GPT-4o.

Starting today, GPT-5 is available for all ChatGPT users. Paid users can enjoy higher usage limits, with Pro users having unlimited access and receiving the enhanced GPT-5 Pro. Free users will switch to the GPT-5 mini model after reaching their usage limit.

Microsoft also announced the integration of GPT-5 into several of its products, including Microsoft 365 Copilot, GitHub Copilot, and Azure AI Foundry.

Three Major Advantages of GPT-5

GPT-5 has achieved significant improvements in three key areas: programming, creative writing, and health consulting.

In programming, GPT-5 is able to create visually appealing responsive websites, applications, and games based solely on a single prompt. It excels in complex front-end generation and debugging large codebases.

In terms of creative writing, GPT-5 is capable of handling complex structural writing tasks, such as unrhymed iambic pentameter or free verse poetry. Nick Turley, Vice President of ChatGPT Business, stated that GPT-5 shows "better taste" in creative tasks and responds more naturally.

In the field of health consulting, GPT-5 can more actively identify potential health issues and help users interpret medical results. In the HealthBench Hard Hallucinations test, GPT-5's error rate for misinformation was only 1.6%, significantly lower than that of previous models.

Security Enhancement

OpenAI has introduced a new security training format called "secure completion." This approach is more nuanced, better able to guide dual-use concerns, enhances robustness against ambiguous intentions, and reduces unnecessary over-rejection.

GPT-5 has been trained to recognize when tasks cannot be completed, avoiding guesswork, and can explain limitations more clearly. This reduces unfounded assertions.

New Chat Preset Personality

OpenAI has launched a research preview version of ChatGPT with four preset personalities: Cynic, Robot, Listener, and Nerd. Users can adjust these personalities at any time in the settings to match their communication style.

These new personalities were initially applicable to text chat and will later expand to voice chat. OpenAI stated that all new personalities meet or exceed the internal evaluation criteria for reducing sycophantic behavior.

Microsoft Complete Integration

Microsoft swiftly announced the integration of GPT-5 into its extensive product line. Microsoft 365 Copilot will leverage GPT-5 to better handle complex issues and understand user context. Consumers can experience GPT-5 through the new intelligent mode of Microsoft Copilot.

Developers will receive GPT-5 support through GitHub Copilot and Visual Studio Code. The Azure AI Foundry platform will provide all GPT-5 models, equipped with AI-driven model routers.

Tests by Microsoft's AI Security team show that GPT-5 exhibits one of the strongest AI security configurations among all models from OpenAI under various attack modes.