OpenAI Meta: ”New AI capable of reasoning and planning will soon be released”

OpenAI and Facebook's parent company Meta are set to launch new artificial intelligence (AI) models capable of reasoning and planning, according to the Financial Times.
The company reported on the 9th (local time). The inference and planning are done by "Artificial General Intelligence" (AGI).
This is a key step towards achieving "Intelligent Intelligence" and is attracting attention.
OpenAI and Meta executives will reportedly launch the next generation version of their Large Language Model (LLM) soon.
Then, in a teaser, the company revealed that the new AI model is capable of reasoning and planning. Being able to reason means that it is not limited to being a simple machine, but has the cognitive capabilities of a human.
Meta said it will release LLAMA3, capable of reasoning and planning, within the next few weeks. Meta has announced that it will open the code to the public.
"Our AI models are the first to be developed," Joel Pino, vice president of AI research at the company, said at an AI Day event in London on Wednesday.
"We're focused on finding ways to actually reason, plan and remember, rather than just interact."
OpenAI, which is backed by Microsoft, also calls it "GPT-5."
"Next-generation GPT will solve hard problems like inference," said Brad Lightcap, chief operating officer at NVIDIA.
"We will see AI that can perform more complex tasks in more sophisticated ways," he said, but noted that reasoning capabilities are still in their early stages.
He noted that existing AI systems are "very good at small, one-off tasks, but their capabilities are still very limited."
Meta's Chief AI Scientist Yann Lekun also spoke at the event, saying, "Current AI systems
"They just produce words one by one without any thought or planning," he said, "and still make silly mistakes because they have difficulty handling complex questions and remembering information for long periods of time."
"But with inference capabilities, AI models can search for possible answers, plan sequences of actions and build mental models."