Meta is launching a new program to adopt the Llama AI model in startups.
The program, the startup, Lama, provides “direct support” from Meta’s llama team and funding in certain cases. The US-based company that was founded raises less than $10 million in funding, has at least one staff developer and is eligible to apply by the May 30th deadline.
“Members can receive up to $6,000 for up to six months to offset the costs of building and enhancing generative AI solutions,” Meta wrote in a blog post. “Our experts will work closely with them to launch and explore advanced use cases for llamas that may benefit the startup.”
The launch of the Llama Startup program is as Meta seeks to solidify its lead in a highly competitive open model space. Tech Giant’s Llama model has so far won over 1 billion downloads, but rivals such as Deepseek, Google and Alibaba’s Qwen have threatened to support Meta’s efforts to establish a wide range of model ecosystems.
Lama has struggled with several set-offs over the past few months because he has not helped the problem.
The Wall Street Journal reported last week that Meta had slowed the deployment of its flagship AI model, the Llama 4 Behemoth, on the decline in performance of models on major benchmarks. In April, Meta had to dodge allegations that he had cheated on the popular crowdsourced AI benchmark LM Arena. The company used the version of the Llama 4 Maverick model “Optimized for Conversation” to achieve a high score on the LM Arena, but has released a different version of Maverick.
Meta has great ambitions for Llama and its broad generator AI portfolio. Last year, the company predicted that its generative AI products would earn between $2 billion and $3 billion in revenue in 2025, and between $1.4 trillion by 2035.
Meta has revenue sharing agreements with some companies that host the Llamas model. The company recently launched an API to customize its Llama release. Meta AI, the AI assistant for Llama-powered AI, could ultimately show ads and offer subscriptions with additional features, CEO Mark Zuckerberg said in the company’s first quarter revenue call.
These products have proven expensive to build. In 2024, Meta’s “genai” budget exceeded $900 million, potentially exceeding $1 billion this year. This does not include the infrastructure required to run and train the model. Meta previously said it plans to spend between $60 billion and $80 billion on capital expenditures in 2025, primarily in new data centers.