Meta aims to release its next AI model, Llama 4.X, by year-end, according to two sources.
The release of Llama 4 models in April faced criticism for underperformance, some developers said.
A team within Meta Superintelligence Labs is now trying to fix bugs within Llama 4, sources say.
Meta is racing to ship its next-generation Llama AI model by the end of the year in one of the first projects to emerge from the newly formed unit Meta Superintelligence Labs.
A team within TBD, one of four groups part of Meta Superintelligence Labs (MSL), is developing Llama 4.X, with the aim of getting the models production-ready in time for the targeted year-end release, according to two people familiar with the matter, who asked to remain anonymous because they were not permitted to speak to the press. Llama 4.X is also interchangeably called Llama 4.5 by some internally, they said.
Meta’s release of its Llama 4 models in April, which includes Scout and Maverick, was met with a flat response from some developers who felt it underdelivered in real-world tasks like coding, reasoning, and following instructions. The TBD team working on Llama 4.X is now also attempting to fix bugs and revive Llama 4, according to the people Business Insider spoke to.
The company was also developing an AI model called Behemoth as part of the Llama 4 family of models. The Wall Street Journal reported in May that Meta postponed the rollout.
A Meta spokesperson declined to comment and pointed to CEO Mark Zuckerberg’s comments about MSL in its second-quarter earnings call in July. Zuckerberg said on the call that the company set up MSL to focus on developing its next generation of AI models.
“We’re making good progress towards Llama 4.1 and 4.2, and in parallel, we’re also working on our next generation of models that will push the frontier in the next year or so,” Zuckerberg said.
Zuckerberg first announced the formation of MSL in an internal memo in June. By August, the company had reorganized its AI teams around four pillars: training, research, product, and infrastructure.
In a follow-up memo that month, MSL head Alexandr Wang explained that a subgroup called TBD would be responsible for “training and scaling large models to achieve superintelligence,” including developing an “omni model.” Wang’s email does not provide further details about “omni.”
The launch of MSL comes after Zuckerberg went on an AI talent hiring spree in recent months, reportedly offering multimillion-dollar compensation packages to top AI researchers from competing labs, including OpenAI and Google DeepMind.
Less than two months after MSL launched, it is already losing some staff. At least eight employees, including researchers, engineers, and a senior product leader, have left the company in the last two months.
Have a tip? Contact Jyoti Mann via email at jmann@businessinsider.com or Signal at jyotimann.11. Use a personal email address and a nonwork device; here’s our guide to sharing information securely.
Read the original article on Business Insider
