On July 24th US time, The Verge reported that earlier this year, Microsoft engineers were preparing server capacity for OpenAI's next-generation GPT-5 model, expected to be released as early as late May. After additional testing and some delays, sources familiar with OpenAI's plans told me that GPT-5 is now expected to be officially released as early as early next month.
OpenAI CEO Sam Altman recently revealed on X (formerly Twitter): "We are about to release GPT-5", and he also previewed some new features on Theo Von's podcast. In the show, Altman experienced GPT-5 for the first time! He was shocked and stunned, sitting paralyzed, that moment was as shocking as seeing an atomic bomb explosion! The response from GPT-5 made him sit paralyzed in his chair, at that moment, feeling how small humans are.
Additionally, he claimed to provide free, all-day GPT-5 for everyone, with another utopian vision - 8 trillion tokens in the future, to be distributed to all 8 billion people on Earth.
Before Altman's appearance, GPT-5 had already been "previewed" in the industry, which intensified external speculation about the imminent release of the new generation model. I understand that OpenAI plans to officially launch GPT-5 in early August, simultaneously releasing mini and nano versions, which will also be accessible through its API.
I contacted OpenAI for comment on GPT-5's August release date, but the company had not responded as of the time of writing.
GPT-5: Not Just an Upgrade, But a Fusion
Altman previously described GPT-5 as a "system that integrates many of our technologies" because it will include o3 series reasoning capabilities, rather than launching a new model separately. This is a major direction OpenAI is advancing: unifying the GPT series and o series models to build a more powerful system and gradually move towards AGI (Artificial General Intelligence).
The AGI declaration is significant for OpenAI - because once OpenAI officially announces achieving AGI, Microsoft will lose its rights to OpenAI's revenue and future AI models. Currently, Microsoft and OpenAI are renegotiating their cooperation agreement, as OpenAI needs Microsoft's approval to convert part of its business to a for-profit company.
However, GPT-5 is reportedly not yet up to AGI standards. Altman has publicly stated that GPT-5 will not have "golden-level capabilities" for several months after its release.
Multiple Versions Launched Simultaneously, Simplified User Experience
Multiple versions of GPT-5 will be launched simultaneously:
- GPT-5 main version (with reasoning capabilities): Will be integrated into ChatGPT and OpenAI API;
- GPT-5 mini version: Also supporting ChatGPT and API;
- GPT-5 nano version: Expected to be provided only through API, not entering ChatGPT.
This unification will also simplify the user experience - no need to struggle with which model to use, directly obtaining the optimal solution through a unified interface.
However, although GPT-5 is currently expected to be released in August, OpenAI's release times are often adjusted due to development issues, server pressure, or competitor model releases. For example, earlier this month, I had warned that an OpenAI "open-source language model" might be delayed, and Altman confirmed the delay a few days later, citing the need for additional safety testing and high-risk review.
First Open-Source Model Also Imminent
Stock Investment Network still hears that OpenAI's open-source language model will be released soon, expected to go online before the end of July, ahead of GPT-5. Sources say this open-source model is functionally similar to "o3 mini" and also has reasoning capabilities. This will be the first open-source weight model released by OpenAI since GPT-2 in 2019, and will be provided through Azure, Hugging Face, and other major cloud platforms.
This article is from the WeChat public account "Stock Investment Network" (ID: tradesmax), author: StockWe.com, published with authorization from 36kr.
