Keep pulling the thread on GPT Live.
OpenAI's GPT Live is built on a full duplex architecture, which allows the model to listen and speak at the same time.
The GPT Live model can delegate tasks like searching or reasoning to another model, such as GPT-5.5 or GPT-5.6, while continuing the live conversation.
According to Artificial Analysis tests, Grok 4.5's cost per task is 31 cents, compared to $1.80 for Opus 4.8 and $2.75 for Fable-5.
Stonk Daddy predicts Groq 4.5 will capture enterprise market share from Chinese open source models by offering superior performance at a similar cost without the associated geopolitical risks.
Ethan Mollick asserts that both GPT-5.6 SOL and Fable-5 represent significant advancements over previous models, creating a large performance gap with the next-best AIs.
According to leaker Leo, OpenAI's GPT-6 is scheduled to launch in approximately one month, possibly in late July, which is earlier than previously expected.
Leaker Leo claims that GPT-6 will be based on a new, significantly larger pre-training dataset compared to the 4 trillion token base used for GPT-5.5 and GPT-5.6.
Leaker Leo reports that Anthropic's Fable 5.1 is in the late stages of its development pipeline and a release is expected in the coming weeks.
Andrew Curran predicts that OpenAI's next release, GPT-6, will arrive much sooner than expected, possibly within the next four weeks.
Andrew Curran predicts that the US government will almost certainly delay the general release of GPT-6, at least initially.
According to Andrew Curran, Elon Musk has a 10 trillion parameter Grok model currently in training.
OpenAI's new voice model, GPT Live, is available in two versions: GPT Live 1 and GPT Live Mini.