According to a report by The Verge, OpenAI might launch GPT-5 in early August. The same information was confirmed by Axios, and aligns with a recent claim by OpenAI CEO Sam Altman that the company is “releasing GPT-5 soon.”
Back in February, Sam Altman said GPT-5 would unify many of OpenAI’s technologies. One of the major changes in the new model may be the addition of reasoning capabilities to the main product. Now, the o3 reasoning model is one of the various separate products that users need to specifically choose.
OpenAI plans to roll out the main GPT-5 along with mini and nano variants. The main combined version and the mini GPT-5 may be available via both ChatGPT and the API, while the nano version could be limited to API access only.
Aside from reducing the complexity of its user product, OpenAI also aims to combine its models to make them more capable. The company’s main goal is to achieve artificial general intelligence, or AGI, which would be a vastly more capable system than the currently available models. The other strongly anticipated model from OpenAI is the company’s open-weight model. After a couple of delays, the open language model might ship before the end of July. OpenAI’s goal is to release it before the launch of GPT-5.
The open language model may have abilities similar to the o3 mini, complete with reasoning capabilities. That would be OpenAI’s first open-weight model since 2019’s release of GPT-2. These types of large-language models are publicly available and can be downloaded and used freely by users. That’s why OpenAI’s open-weight model is expected to be available on cloud providers such as Hugging Face and Azure.
People in AI tend to hype up their achievements a little too much, but I’m quite excited about OpenAI’s new model. Despite fierce competition from Google’s Gemini, Anthropic’s Claude, Perplexity AI, and others, ChatGPT remains the most popular and, in my opinion, the best AI product.
Even now, ChatGPT is capable of some impressive feats, so any improvement would make it even better. My main issue with all AI chatbots is that they hallucinate, which is why they often generate misleading information. GPT-5 is unlikely to solve that issue, but if it manages to sound less confident when hallucinating, it’d be a step in the right direction.
