Reka AI, a startup specializing in multimodal large language models, has announced the completion of a new funding round backed by major investors Nvidia and Snowflake. While the exact funding amount and new valuation were not disclosed, reports indicate the deal has tripled the company´s valuation compared to its previous round. Reka, founded in 2022, is focused on developing large-scale Artificial Intelligence models capable of analyzing video, images, text, and audio using ultra-efficient training and inference techniques to reduce operational costs.
The company’s product suite includes Reka Flash, a flagship multimodal model, as well as Reka Vision and Reka Research, which have moved into general availability, offering advanced features such as complex question answering and visual search. These capabilities are designed for both enterprise and research use, underlining Reka´s drive towards accessible, scalable Artificial Intelligence. Notably, Reka Flash 3.1 reportedly rivals leading models like Quen3-32B and OpenAI´s o3-mini, while maintaining efficiency thanks to Reka’s proprietary quantization tool, Reka Quant. This technology allows almost lossless reduction in model size, enabling robust performance on resource-constrained devices with minimal accuracy tradeoff.
Snowflake, a recurring collaborator and former suitor, maintains close strategic ties, planning to integrate Reka’s models and services more deeply in its platform. Vivek Raghunathan, Snowflake’s vice president of Artificial Intelligence engineering, praised Reka for competing at a level occupied by an elite group of tech giants and researchers. Reka’s team has grown significantly over the past year, and the company intends to use its new capital to accelerate technical development and further scale its multimodal infrastructure for broader enterprise adoption and global reach.
