What is a Virtual Private Server (VPS) and Wh
As the demand for flexible, affordable, and scalable ho...
Sam Altman announced on Wednesday that OpenAI is about to release a new AI model called GPT-4.5, which will be OpenAI's last "non-chain-of-thought model." He also revealed that GPT-5 is expected to be released in the coming months and will be free to use!
Related Reading:
Complete Guide: Deploying DeepSeek R1 on an RTX 4090 GPU
DeepSeek AI: China's Rising Star in the Global AI Race
Host Your DeepSeek R1 with SurferCloud RTX 4090 GPU Servers
SurferCloud GPU UHost: Deploy Your DeepSeek R1 with Ollama
How to Apply for Free Trial of DeepSeek R1 on SurferCloud UModelVerse
UModelVerse Launches with Free Access to deepseek-ai/DeepSeek-R1
According to Altman, OpenAI is close to launching GPT-4.5, internally referred to as "Orion." The model is expected to be released within weeks.
Analysts believe the release of GPT-4.5 signifies the end of an era. Altman stated that this will be the last "non-chain-of-thought model" OpenAI develops—meaning it does not rely on additional computational resources to simulate human reasoning. Unlike OpenAI’s o-series models (such as o3), which adopt a reasoning-driven approach, non-chain-of-thought models often struggle in fields like mathematics and physics.
Altman also mentioned that OpenAI plans to merge the GPT series (the core technology behind ChatGPT) with the o-series models, creating an integrated system capable of using all tools while deciding when to engage in deep thinking or provide quick responses. The goal is to simplify user experience and eliminate the need for users to select from complex model options.
Currently, users who want ChatGPT to use a specific model (e.g., o3-mini) must click the "Reason" button in the prompt bar or manually select models from a dropdown menu. Altman expressed dissatisfaction with this system, saying:
"We want AI to run seamlessly. We, like you, hate the 'model selector.'"
Altman also revealed that OpenAI is accelerating the development of GPT-5, with a planned release in the coming months. GPT-5 will integrate multiple AI technologies from OpenAI, including o3, which means there will no longer be a standalone o3 model release.
Once GPT-5 is officially launched, OpenAI will provide free users with unlimited access to a 'standard intelligence level'—though this will still be subject to "abuse limits." (Altman did not disclose specific details about these limits.)
Additionally, ChatGPT Plus subscribers will gain access to higher intelligence levels of GPT-5, while ChatGPT Pro users will have access to even more advanced capabilities.
Altman stated:
"The model will integrate voice, canvas, search, deep research, and more. One of our core goals is to unify all models into a system that seamlessly integrates all tools, knows when to engage in deep reasoning, and can handle a wide variety of tasks."
Despite the excitement, GPT-4.5 (Orion) has faced multiple performance challenges and technical issues during development. Some reports indicate that the improvement from GPT-4 to GPT-4.5 is not as significant as the leap from GPT-3 to GPT-4, raising concerns about whether GPT-4.5 offers a substantial upgrade over GPT-4o.
These announcements shocked the AI community, prompting many to ask when the new models will be officially unveiled. Altman responded that releases will happen in weeks and months, not years.
Some users speculated that OpenAI’s decision to offer GPT-5 for free indicates pressure from China’s DeepSeek AI—suggesting that OpenAI may be trying to retain users before they switch to competing platforms.
Others expressed excitement and gratitude for being able to use GPT-5 for free.
Some users also proposed new feature ideas:
"Can you design a 'health bar' to show how many questions we have left and a 'brainpower bar' for reasoning-based challenges? It would be fun to turn AI into a game!"
This news signals a major shift in OpenAI's approach, and the AI industry will be watching closely to see how GPT-4.5, GPT-5, and DeepSeek shape the future of AI applications.
As the demand for flexible, affordable, and scalable ho...
Monitoring the performance of your Content Delivery Net...
GPU servers are revolutionizing the field of deep learn...