New development suggests that GPT-5 might debut in August, while OpenAI's open-source language model could potentially surface ahead of this timeline.
OpenAI, the leading AI research organisation, is preparing to launch a new open-weight language model before the end of July. This marks the company's first open-weight model release since GPT-2 in 2019.
The new model is not exclusive to OpenAI's API like the nano version of GPT-5. Instead, it will be available on platforms like Azure, Hugging Face, and other cloud services, making it accessible to a wider audience.
Key Features of the New Model
The new open model is designed to offer improved reasoning skills, setting it apart from previous models. It will dynamically route queries to the appropriate subsystem, providing integrated capabilities like memory, reasoning, vision, and task completion all in one system.
This model is similar to the "o3 mini" and is expected to enhance coding and reasoning performance, including better handling of complex programming tasks such as updating large outdated codebases. It is positioned as an AI “system” rather than a single model, designed to manage multiple tools and diverse tasks seamlessly.
Versions of GPT-5
GPT-5, OpenAI's next big AI model, is expected to launch as early as August 2025, and will come in three sizes: standard, mini, and nano.
The standard version of GPT-5 will be available on both ChatGPT and OpenAI's API, offering a versatile solution for users and developers alike. The mini version, similar to the standard version, is also likely to be offered on ChatGPT and OpenAI's API, providing a smaller, more efficient variant.
The nano version of GPT-5 is said to be available through OpenAI's API only, targeting developers or applications needing a smaller footprint.
Integration and Backing
The model is Microsoft-backed and integrates OpenAI’s o3 model along with other technologies, signaling deeper collaboration with Microsoft’s cloud and AI infrastructure.
Release Timing
Though no official confirmation has been made, reports and insider sources suggest the release could happen anytime in August 2025, with potential shifts depending on development or external factors.
In a podcast appearance, OpenAI CEO Sam Altman revealed that GPT-5 answered a question he couldn't answer, hinting at the model's advanced capabilities. The release of this new open model precedes the launch of GPT-5, offering a sneak peek into the future of AI.
The new open model is not integrated with ChatGPT, but it serves as an exciting step towards the anticipated launch of GPT-5 and its integrated reasoning capabilities. Stay tuned for more updates on these groundbreaking AI developments.
References:
[1] OpenAI Blog: Announcing GPT-5
[2] TechCrunch: OpenAI's GPT-5 will be a unified AI system with three sizes
[3] The Verge: OpenAI's GPT-5 will be a unified AI system with three sizes
The new model, an open-weight language model about to be launched by OpenAI, will feature artificial-intelligence capabilities, having integrated capabilities like memory, reasoning, vision, and task completion all in one system. This model, not exclusive to OpenAI's API, is also planned to be available on platforms like Azure, Hugging Face, and other cloud services, set to be more accessible to a wider audience.
The upcoming GPT-5, OpenAI's next big AI model, will reportedly come in three sizes: standard, mini, and nano, with the standard and mini versions expected to be available on both ChatGPT and OpenAI's API. While the new open model is not integrated with ChatGPT yet, it signifies an exciting step towards the anticipated launch of GPT-5, which is expected to have advanced artificial-intelligence capabilities.
