This year's most anticipated product from OpenAI has arrived.

On Thursday, August 7 (Eastern Time), OpenAI announced the launch of its new flagship artificial intelligence (AI) model, GPT-5. It is OpenAI's first "all-in-one" AI system, marking the first time OpenAI has combined the reasoning capabilities of its o-series models with the fast response capabilities of the GPT series.

OpenAI CEO Sam Altman spoke highly of GPT-5 at the new model launch event, calling it "the best model in the world" and a "major upgrade" over previous models. He stated that its release signifies an "important step" for OpenAI on the path to achieving artificial general intelligence (AGI).

OpenAI introduced that GPT-5 has performed exceptionally well in multiple benchmark tests, reaching cutting-edge levels in fields such as programming, mathematics, and healthcare. GPT-5 achieved an accuracy rate of 74.9% in the SWE-bench Verified coding test, slightly surpassing Anthropic's newly released model Claude Opus 4.1, which was launched on Tuesday. Additionally, GPT-5's hallucination issue has been significantly improved, with an error rate of only 4.8%, far lower than the 20.6% of its predecessor, GPT-4o.

Starting from Thursday, GPT-5 has been made available to all free ChatGPT users as well as paid subscribers of Plus, Pro, and Team plans as the default model. It will be launched on Enterprise and Edu paid plans within a week.

Similar to GPT-4o, the difference between the free and paid versions of GPT-5 lies in usage limits. Plus users enjoy higher usage quotas, while Pro users can use it unlimitedly and access the enhanced version, GPT-5 Pro. For free users, the full reasoning capabilities may take a few days to be fully rolled out. Once free users reach GPT-5's usage limit, OpenAI will switch them to the smaller model, GPT-5 mini.

OpenAI also stated on Wednesday that it will provide ChatGPT products to U.S. federal government agencies at a symbolic annual fee of $1. Specifically, this refers to the enterprise version of ChatGPT, which includes enhanced security and privacy features.

Microsoft Integrates GPT-5 Immediately After Announcement

Shortly after OpenAI officially announced GPT-5, Microsoft announced that starting from Thursday, it will integrate GPT-5 into its extensive product portfolio, including platforms such as 365 Copilot, Copilot, GitHub Copilot, and Azure AI Foundry. This allows Microsoft's enterprise and consumer users to immediately experience GPT-5's advanced reasoning capabilities and programming advantages.

GPT-5 Boasts Three Key Advantages in Programming, Creative Writing, and Healthcare

OpenAI's GPT-5 release announcement begins by stating that GPT-5 is OpenAI's "smartest, fastest, and most practical model, with built-in thinking capabilities that enable everyone to possess expert-level wisdom."

According to OpenAI, as its "most powerful model," GPT-5 has achieved significant improvements in three key areas.

First is programming capability. GPT-5 is OpenAI's most powerful coding model to date, excelling in complex frontend generation and large codebase debugging. It can create aesthetically pleasing and responsive websites, applications, and games with just a single prompt. Early testers noted improvements in design choices such as spacing, typography, and whitespace.

In the SWE-bench Verified benchmark test, which draws real-world coding tasks from GitHub, GPT-5 achieved an accuracy rate of 74.9% on its first attempt after thinking, higher than OpenAI's reasoning model o3 (69.1%) and GPT-4o (30.8%).

Comments point out that this means GPT-5's performance is slightly better than Anthropic's Claude Opus 4.1, launched on Tuesday, and Google DeepMind's Gemini 2.5 Pro, which scored 74.5% and 59.6% in the SWE-bench Verified test, respectively.

However, in the Humanity's Last Exam test, which measures model performance in mathematics, humanities, and natural sciences at an expert level, the enhanced version of GPT-5 with extended reasoning capabilities, GPT-5 Pro, scored 42% when using tools. This is slightly lower than xAI's model Grok 4 Heavy, which scored 44.4%.

Altman stated that GPT-5 is particularly adept at launching entire software applications on demand, known as "ambient coding"—using AI to generate functional code based on natural language prompts, thereby accelerating development speed.

As an example, OpenAI researchers demonstrated asking GPT-5 to create a web app to help English-speaking users learn French, with requirements for an engaging theme, including flashcards, quizzes, a classic Snake game, and a way to track daily learning progress.

The researchers submitted the same prompt to two GPT-5 windows, and two different apps were generated within minutes. OpenAI officials noted that these apps "have some flaws," but users can adjust the AI-generated software according to personal preferences, such as changing the background or adding more tabs.

In terms of creative writing, GPT-5 can handle structurally complex writing tasks, such as unrhymed iambic pentameter or naturally flowing free verse. Nick Turley, Vice President of ChatGPT Business at OpenAI, stated that GPT-5 demonstrates "better taste" in creative tasks, with more natural responses.

Health consultation is the third area of significant improvement.

GPT-5 can more actively flag potential health issues and help users interpret medical results, although OpenAI emphasizes that ChatGPT cannot replace medical professionals.

In the test called HealthBench Hard Hallucinations, GPT-5 with thinking capabilities had a hallucination error rate of only 1.6%. This is far lower than the GPT-4o and o3 models, which had error rates of 15.8% and 12.9%, respectively.

Significantly Reduced Hallucination Probability and New Safety Training Model

OpenAI stated that GPT-5 is more reliable and practical than previous models, providing more accurate answers to real-world questions with a significantly reduced probability of hallucinations.

After enabling web search for anonymous prompts representing ChatGPT's production traffic, GPT-5's responses contained factual errors about 45% less often than GPT-4o; after thinking, GPT-5's responses contained factual errors about 80% less often than o3. As shown in the figure below, GPT-5's response error rate is only 4.8%, compared to 20.6% for GPT-4o and 22% for o3.

OpenAI also stated that it has introduced a new form of safety training for GPT-5, called "safe completions." It teaches the model to provide the most helpful answers possible within safe boundaries. Sometimes, this may mean partially answering the user's question or only providing high-level responses.

If a refusal is necessary, the trained GPT-5 will transparently inform the user of the reason for the refusal and offer safe alternatives.

In controlled experiments and OpenAI's production models, OpenAI found that this safe completion method is more nuanced, better able to guide dual-use questions, enhance robustness to ambiguous intentions, and reduce unnecessary over-refusals.

Michelle Pokrass, Head of Post-Training at OpenAI, stated: "GPT-5 has been trained to recognize when a task cannot be completed, avoid guessing, and explain limitations more clearly, reducing unfounded assertions compared to previous models."

Launch of Four Optional ChatGPT Chat Preset Personalities

OpenAI stated that GPT-5 has improved in instruction execution, with corresponding enhancements in its ability to execute custom instructions. OpenAI will launch a new research preview with four preset personalities for all ChatGPT users.

The initial four personality options—Cynic, Robot, Listener, and Nerd—are all optional, and users can adjust them at any time in the settings to match ChatGPT's communication style with their own.

These four personalities initially apply to text chat and will later be extended to voice chat, allowing users to set ChatGPT's interaction style—whether concise and professional, thoughtful and supportive, or slightly sarcastic—without writing custom prompts.

OpenAI stated that all these new personalities meet or exceed its internal evaluation criteria for reducing obsequious behavior.

Altman Praises Historic Breakthrough, Calls GPT-4 in Comparison

At Thursday's briefing, Altman highly praised GPT-5, positioning it as an important milestone on the path to AGI. He stated:

"At any previous point in history, having something like GPT-5 would have been unimaginable." "This is the first time it feels like talking to an expert in any field."

At the briefing, Altman even went so far as to "disparage" GPT-4 to elevate GPT-5, saying:

"I tried going back to GPT-4, but the effect was quite bad."

GPT-5 adopts a unified system architecture with a real-time router, which can automatically decide whether to respond quickly or engage in in-depth "thinking" based on the type, complexity, and tool requirements of the conversation. This eliminates the need for users to choose appropriate settings, making ChatGPT easier to use.

In internal benchmarks for economically valuable work, GPT-5 in reasoning mode was comparable to or better than expert-level performance in approximately half of the cases, covering over 40 professions such as law, logistics, sales, and engineering. OpenAI VP Nick Turley said: "This model really feels great."

Altman compared using GPT-5 to having a team of experts all with doctoral degrees at one's disposal. He also said: "In many new fields, people are limited by ideas but lack the ability to execute them."

Microsoft's Comprehensive Integration to Seize the Initiative

Microsoft announced on the day of GPT-5's release that it will integrate it into its extensive product line. In terms of enterprise applications, Microsoft 365 Copilot will use GPT-5 to better handle complex issues, maintain focus in long conversations, and understand user context. Enterprise users can process emails, documents, and files through reasoning functions.

For consumers, Microsoft Copilot's new smart mode will utilize GPT-5 to help users find the best solutions. Users can experience GPT-5 for free through copilot.microsoft.com or the Copilot app on Windows, Mac, Android, and iOS devices.

Developers will receive GPT-5 support through GitHub Copilot and Visual Studio Code for writing, testing, and deploying code. The Azure AI Foundry platform will offer all GPT-5 models, equipped with an AI-driven model router to select the optimal model based on the complexity, performance requirements, and cost efficiency of each task.

Microsoft's AI red team tested the GPT-5 reasoning model using strict security protocols, and the results showed that the model demonstrates one of the strongest AI security configurations among OpenAI's 历代 models in various attack scenarios such as malware generation and fraud automation.

This year's most anticipated product from OpenAI has arrived.

On Thursday, August 7 (Eastern Time), OpenAI announced the launch of its new flagship artificial intelligence (AI) model, GPT-5. It is OpenAI's first "all-in-one" AI system, marking the first time OpenAI has combined the reasoning capabilities of its o-series models with the fast response capabilities of the GPT series.

OpenAI CEO Sam Altman spoke highly of GPT-5 at the new model launch event, calling it "the best model in the world" and a "major upgrade" over previous models. He stated that its release signifies an "important step" for OpenAI on the path to achieving artificial general intelligence (AGI).

OpenAI introduced that GPT-5 has performed exceptionally well in multiple benchmark tests, reaching cutting-edge levels in fields such as programming, mathematics, and healthcare. GPT-5 achieved an accuracy rate of 74.9% in the SWE-bench Verified coding test, slightly surpassing Anthropic's newly released model Claude Opus 4.1, which was launched on Tuesday. Additionally, GPT-5's hallucination issue has been significantly improved, with an error rate of only 4.8%, far lower than the 20.6% of its predecessor, GPT-4o.

Starting from Thursday, GPT-5 has been made available to all free ChatGPT users as well as paid subscribers of Plus, Pro, and Team plans as the default model. It will be launched on Enterprise and Edu paid plans within a week.

Similar to GPT-4o, the difference between the free and paid versions of GPT-5 lies in usage limits. Plus users enjoy higher usage quotas, while Pro users can use it unlimitedly and access the enhanced version, GPT-5 Pro. For free users, the full reasoning capabilities may take a few days to be fully rolled out. Once free users reach GPT-5's usage limit, OpenAI will switch them to the smaller model, GPT-5 mini.

OpenAI also stated on Wednesday that it will provide ChatGPT products to U.S. federal government agencies at a symbolic annual fee of $1. Specifically, this refers to the enterprise version of ChatGPT, which includes enhanced security and privacy features.

Microsoft Integrates GPT-5 Immediately After Announcement

Shortly after OpenAI officially announced GPT-5, Microsoft announced that starting from Thursday, it will integrate GPT-5 into its extensive product portfolio, including platforms such as 365 Copilot, Copilot, GitHub Copilot, and Azure AI Foundry. This allows Microsoft's enterprise and consumer users to immediately experience GPT-5's advanced reasoning capabilities and programming advantages.

GPT-5 Boasts Three Key Advantages in Programming, Creative Writing, and Healthcare

OpenAI's GPT-5 release announcement begins by stating that GPT-5 is OpenAI's "smartest, fastest, and most practical model, with built-in thinking capabilities that enable everyone to possess expert-level wisdom."

According to OpenAI, as its "most powerful model," GPT-5 has achieved significant improvements in three key areas.

First is programming capability. GPT-5 is OpenAI's most powerful coding model to date, excelling in complex frontend generation and large codebase debugging. It can create aesthetically pleasing and responsive websites, applications, and games with just a single prompt. Early testers noted improvements in design choices such as spacing, typography, and whitespace.

In the SWE-bench Verified benchmark test, which draws real-world coding tasks from GitHub, GPT-5 achieved an accuracy rate of 74.9% on its first attempt after thinking, higher than OpenAI's reasoning model o3 (69.1%) and GPT-4o (30.8%).

Comments point out that this means GPT-5's performance is slightly better than Anthropic's Claude Opus 4.1, launched on Tuesday, and Google DeepMind's Gemini 2.5 Pro, which scored 74.5% and 59.6% in the SWE-bench Verified test, respectively.

However, in the Humanity's Last Exam test, which measures model performance in mathematics, humanities, and natural sciences at an expert level, the enhanced version of GPT-5 with extended reasoning capabilities, GPT-5 Pro, scored 42% when using tools. This is slightly lower than xAI's model Grok 4 Heavy, which scored 44.4%.

Altman stated that GPT-5 is particularly adept at launching entire software applications on demand, known as "ambient coding"—using AI to generate functional code based on natural language prompts, thereby accelerating development speed.

As an example, OpenAI researchers demonstrated asking GPT-5 to create a web app to help English-speaking users learn French, with requirements for an engaging theme, including flashcards, quizzes, a classic Snake game, and a way to track daily learning progress.

The researchers submitted the same prompt to two GPT-5 windows, and two different apps were generated within minutes. OpenAI officials noted that these apps "have some flaws," but users can adjust the AI-generated software according to personal preferences, such as changing the background or adding more tabs.

In terms of creative writing, GPT-5 can handle structurally complex writing tasks, such as unrhymed iambic pentameter or naturally flowing free verse. Nick Turley, Vice President of ChatGPT Business at OpenAI, stated that GPT-5 demonstrates "better taste" in creative tasks, with more natural responses.

Health consultation is the third area of significant improvement.

GPT-5 can more actively flag potential health issues and help users interpret medical results, although OpenAI emphasizes that ChatGPT cannot replace medical professionals.

In the test called HealthBench Hard Hallucinations, GPT-5 with thinking capabilities had a hallucination error rate of only 1.6%. This is far lower than the GPT-4o and o3 models, which had error rates of 15.8% and 12.9%, respectively.

Significantly Reduced Hallucination Probability and New Safety Training Model

OpenAI stated that GPT-5 is more reliable and practical than previous models, providing more accurate answers to real-world questions with a significantly reduced probability of hallucinations.

After enabling web search for anonymous prompts representing ChatGPT's production traffic, GPT-5's responses contained factual errors about 45% less often than GPT-4o; after thinking, GPT-5's responses contained factual errors about 80% less often than o3. As shown in the figure below, GPT-5's response error rate is only 4.8%, compared to 20.6% for GPT-4o and 22% for o3.

OpenAI also stated that it has introduced a new form of safety training for GPT-5, called "safe completions." It teaches the model to provide the most helpful answers possible within safe boundaries. Sometimes, this may mean partially answering the user's question or only providing high-level responses.

If a refusal is necessary, the trained GPT-5 will transparently inform the user of the reason for the refusal and offer safe alternatives.

In controlled experiments and OpenAI's production models, OpenAI found that this safe completion method is more nuanced, better able to guide dual-use questions, enhance robustness to ambiguous intentions, and reduce unnecessary over-refusals.

Michelle Pokrass, Head of Post-Training at OpenAI, stated: "GPT-5 has been trained to recognize when a task cannot be completed, avoid guessing, and explain limitations more clearly, reducing unfounded assertions compared to previous models."

Launch of Four Optional ChatGPT Chat Preset Personalities

OpenAI stated that GPT-5 has improved in instruction execution, with corresponding enhancements in its ability to execute custom instructions. OpenAI will launch a new research preview with four preset personalities for all ChatGPT users.

The initial four personality options—Cynic, Robot, Listener, and Nerd—are all optional, and users can adjust them at any time in the settings to match ChatGPT's communication style with their own.

These four personalities initially apply to text chat and will later be extended to voice chat, allowing users to set ChatGPT's interaction style—whether concise and professional, thoughtful and supportive, or slightly sarcastic—without writing custom prompts.

OpenAI stated that all these new personalities meet or exceed its internal evaluation criteria for reducing obsequious behavior.

Altman Praises Historic Breakthrough, Calls GPT-4 in Comparison

At Thursday's briefing, Altman highly praised GPT-5, positioning it as an important milestone on the path to AGI. He stated:

"At any previous point in history, having something like GPT-5 would have been unimaginable." "This is the first time it feels like talking to an expert in any field."

At the briefing, Altman even went so far as to "disparage" GPT-4 to elevate GPT-5, saying:

"I tried going back to GPT-4, but the effect was quite bad."

GPT-5 adopts a unified system architecture with a real-time router, which can automatically decide whether to respond quickly or engage in in-depth "thinking" based on the type, complexity, and tool requirements of the conversation. This eliminates the need for users to choose appropriate settings, making ChatGPT easier to use.

In internal benchmarks for economically valuable work, GPT-5 in reasoning mode was comparable to or better than expert-level performance in approximately half of the cases, covering over 40 professions such as law, logistics, sales, and engineering. OpenAI VP Nick Turley said: "This model really feels great."

Altman compared using GPT-5 to having a team of experts all with doctoral degrees at one's disposal. He also said: "In many new fields, people are limited by ideas but lack the ability to execute them."

Microsoft's Comprehensive Integration to Seize the Initiative

Microsoft announced on the day of GPT-5's release that it will integrate it into its extensive product line. In terms of enterprise applications, Microsoft 365 Copilot will use GPT-5 to better handle complex issues, maintain focus in long conversations, and understand user context. Enterprise users can process emails, documents, and files through reasoning functions.

For consumers, Microsoft Copilot's new smart mode will utilize GPT-5 to help users find the best solutions. Users can experience GPT-5 for free through copilot.microsoft.com or the Copilot app on Windows, Mac, Android, and iOS devices.

Developers will receive GPT-5 support through GitHub Copilot and Visual Studio Code for writing, testing, and deploying code. The Azure AI Foundry platform will offer all GPT-5 models, equipped with an AI-driven model router to select the optimal model based on the complexity, performance requirements, and cost efficiency of each task.

Microsoft's AI red team tested the GPT-5 reasoning model using strict security protocols, and the results showed that the model demonstrates one of the strongest AI security configurations among OpenAI's 历代 models in various attack scenarios such as malware generation and fraud automation.