GPT-6

On April 7th, the GPT-6, internal designer “Spud” (Toast), was developed in secret for two years and completed pre-training at the Stargate data centre in Texas on March 24th. In a podcast interview, the CEO of OpenAI Greg Brockman personally confirmed the existence of the model as "not incremental improvement, but rather a major change in how we think about model development". On the same day, Sam Altman stated to all employees that it was "a very powerful model that could really accelerate economic development". According to the leak..

On April 6th, while people were anxiously waiting for DeepSeek V4, US AI did not stop, and OpenAI launched the Red Alert at the end of last year, focusing on building a super-powerful AI, possibly the next generation of GPT-6, the latest of which was scheduled for April 14th. Netizen iruletheworldmo reports that information from the OpenAI team indicates that GPT-6 is due on April 14, and pre-training has been completed on March 17, and post-training..

According to October 19, OpenAI does not plan to launch GPT-6 this year, but this does not mean that the company will not publish a new model. Currently, multiple models have been launched in the OpenAI GPT-5 series. The default version is GPT-5 Auto, which allows automatic switching between standard and reasoning models. The reasoning model allows for longer “thinking” to generate better-quality answers, while the GPT-5-instant model is based on a non-critic model, with a faster response. Auto Mode for GPT-5

August 20, OpenAI CEO Sam Altman revealed the company's future product development direction to the media in San Francisco last week. Although GPT-5 has just been launched, Altman has made it clear that GPT-6 is already on the way, and the release time will be shorter than the interval between GPT-4 and GPT-5. According to Altman, GPT-6 will be a breakthrough in several ways. It will not only be able to respond to user commands, but will also be able to adapt according to the user's personal preferences, habits and characteristics...

March 1, 2011 - Tech media outlet smartprix published a blog post yesterday (February 28) reporting that OpenAI accidentally leaked the number of GPUs that may be needed for GPT-6 training in a video introducing the GPT-4.5 model, suggesting that it will be much larger than ever before. Note: At the 2:26 mark of the GPT-4.5 model introduction video, the chat transcript of OpenAI's demonstration of GPT 4.5's capabilities includes the phrase "Num GPUs for GPT 6 Training"...