Cerebras said on 29 April 2025 that Meta had teamed up with Cerebras to offer ultra-fast inference in Meta’s new Llama API.
Cerebras said developers building on the Llama 4 Cerebras model in the API can expect generation speeds up to 18 times faster than traditional GPU-based solutions. It named low-latency voice, interactive code generation, multi-step reasoning and real-time agents as use cases.
Developers can choose Cerebras from the model options within the Llama API.
Cerebras said serving Llama models from Meta’s API gives it exposure to a wider global developer audience. It cited Artificial Analysis figures of over 2,600 tokens per second for Llama 4 Scout.
“Developers building agentic and real-time apps need speed,” said Andrew Feldman, CEO and co-founder of Cerebras.