SambaNova Systems
September 10, 2024
SambaNova Systems Unicorn News - September 10, 2024
SambaNova Systems has launched its new inference cloud featuring Meta's Llama 3.1 405B model. The service offers both free and paid enterprise tiers, with a developer tier planned for later this year. This launch positions SambaNova as a key player in the AI infrastructure space.
() Introduction
() SambaNova Launches Inference Cloud with Llama 3.1 405B Model
SambaNova Systems has unveiled its new inference cloud, which incorporates Meta's Llama 3.1 405B model. This advanced AI model can generate tokens at a rate of 132 per second, establishing SambaNova as a key player in the AI infrastructure sector. The company's SN40L-based systems reportedly operate at nearly twice the speed of the fastest GPU systems, based on data from Artificial Analysis. The cloud service is now available in both free and paid enterprise tiers, with a developer tier expected to launch later this year. This service can process over 100 tokens per second, highlighting a significant improvement in AI model serving capabilities.
