TractoAI is a serverless platform for running data and compute-intensive workloads at scale. Data scientists, AI/ML engineers, and developers can set up and scale AI and data workflows to hundreds of GPUs in seconds.
We’re excited to announce the launch of TractoAI private beta program, designed for innovators and builders in the AI/ML space. Our platform equips engineers and startups with the tools they need to prepare datasets, train models, and launch production-grade AI apps — all with efficiency and scale in mind.
This launch comes on the heels of a major milestone achieved by one of our early adopters, Pleias, a pioneering French AI research lab. Pleias recently unveiled the world’s first suite of large language models trained exclusively on open, permissively licensed data. Their 1-billion-parameter model, Pleias-Nano, was trained on the TractoAI platform, demonstrating the platform’s ability to handle complex AI challenges while delivering exceptional performance.
"TractoAI reliability and speed allowed us to pretrain 1B text LLM on 15TB of unstructured data. Integrating data preprocessing and training in one platform saved our ML team hours of work and kept us on schedule" shared Pierre-Carl Langlais, Co-founder at Pleias.
We are extending an invitation to AI teams building next Gen AI services/apps to join our next design partner cohort. We are selecting 4 partners for whom will build out AI backend over the next 3 months. Learn more and apply here.