Together AI
AI cloud infrastructure provider
About
Founded in June 2022 by Vipul Ved Prakash, Ce Zhang, Percy Liang, and Chris Re in San Francisco, Together AI operates the AI Acceleration Cloud, a full-stack platform for production AI covering inference, training, fine-tuning, agentic workflows, and synthetic data generation. The platform serves 450,000+ developers and enterprises including Salesforce, Zoom, SK Telecom, The Washington Post, Cognition, and Zomato. The company has 200 MW of secured power capacity and is deploying clusters of NVIDIA Blackwell GPUs across North American data centers, including a partnership with Hypertec for 36,000 NVIDIA GB200 NVL72 GPUs.
Products
-
FlashAttention-4
— An optimized attention mechanism delivering up to 1.3× faster performance on NVIDIA Blackwell GPUs.
Customer sectors: ai researchers, ml engineers
-
ATLAS
— Runtime-learning accelerators delivering up to 4x faster large language model inference.
Customer sectors: ml engineers, ai developers
-
Together GPU Clusters
— Self-service, scalable NVIDIA GPU clusters optimized for AI workloads.
Customer sectors: ai researchers, ml engineers, data scientists
-
Batch Inference API
— API to cost-effectively process billions of tokens asynchronously at 50% lower cost for most models.
Customer sectors: ml engineers, data scientists, enterprise ai teams
-
Fine-Tuning Platform
— Platform to fine-tune open-source models with latest techniques without managing training infrastructure.
Customer sectors: ml engineers, ai researchers, enterprise ai teams
Market research & competitive positioning
Source-backed TAM / SAM / SOM sizing, displacement analysis, and an interactive peer positioning map for Together AI — available to members.
Related Companies
Other startups in Developer Tools & Infrastructure
