OpenAI has released the next generation of its models, GPT-5, which the company claims is the most powerful AI among existing models.

This AI includes various models; the main one among them is GPT-5, which is a strong reasoning model for performing various tasks ranging from generating high-quality texts to coding, and it has a Context Window of over 400,000 tokens. When users exceed their limit for this model, the GPT-5 Mini model becomes available to them, which is a weaker model but still stronger than previous Mini models. Additionally, there is GPT-5 Nano, the lightest model among these, which is only accessible via API, and finally GPT-5 Pro, considered the most powerful model in this family and available only to Pro subscribers.

However, this time there is no long model selection menu that confused many users, especially beginners in ChatGPT; instead, all models are summarized into two options: GPT-5 and GPT-5 Thinking. In this way, ChatGPT automatically decides which model to use based on the request you make, allocating fewer resources and responding faster to simple questions, and dedicating more resources and thinking more for complex questions. For those who need the model to think more, they can manually select GPT-5 Thinking or tell the model to think more.

In practice, according to various tests conducted by the company, GPT-5 is considered a top-tier model, and even if it is not first in a test, its gap with competitors has greatly narrowed. In programming and in SWE-Bench and Aider Polyglot tests, it has scored higher than competitors including Claude.

This model also makes fewer mistakes than previous models, hallucinates less, and especially when the user's question violates the rules applied to it, instead of completely avoiding answering, it tries to respond within the limits of the rules.

Moreover, although users should not rely heavily on these models for interpreting their medical test results, many people do so anyway, so the company has emphasized the improved accuracy and precision of this model compared to previous ones.

This model is now available to all users, including free users, who can send 10 messages to GPT-5 every 5 hours for free, with subsequent messages answered by the Mini model. Plus subscribers can send 80 messages every three hours with GPT-5 and 200 messages per week with GPT-5 Thinking.

🌐📲 @TechnologyBoxs