⚡️ FlashAttention-4: up to 1.3× faster than cuDNN on NVIDIA Blackwell →

Introducing Together AI's new look →

🔎 ATLAS: runtime-learning accelerators delivering up to 4x faster LLM inference →

⚡ Together GPU Clusters: self-service NVIDIA GPUs, now generally available →

📦 Batch Inference API: Process billions of tokens at 50% lower cost for most models →

🪛 Fine-Tuning Platform Upgrades: Larger Models, Longer Contexts →

Request your B200 cluster

Reserve large-scale NVIDIA B200 clusters, turbocharged by our team of AI researchers.

  • NVIDIA HGX B200 with 192GB HBM3e — Blackwell architecture for AI reasoning.
  • Clusters from 256 → 2,000+ GPUs. On-demand from $5.50/hr, reserved from $4.00/hr.
  • Up to 90% faster BF16 training vs optimized Hopper with Together Kernel Collection.
  • Managed Kubernetes or Slurm orchestration — deploy in minutes.
  • Available across 25+ cities in North America, Europe, and Asia.
Trusted by

Large scale NVIDIA GPU clusters, custom built for you

As an NVIDIA Cloud Partner, we have massive clusters ready for you right now, and can also work with you to build GPU Clusters specific to your project needs.

  • NVIDIA GB200 NVL72

    A rack-scale, liquid-cooled supercomputer that enables 72 NVIDIA Blackwell GPUs to act as one massive GPU, delivering 1.4 exaFLOPs of AI performance and up to 30TB of fast memory.

    Learn more
  • NVIDIA Blackwell GPU

    Delivering up to 15X more real-time inference and 3X faster training to accelerate trillion-parameter language models compared to the NVIDIA Hopper architecture generation.

    Learn more
  • NVIDIA HGX H200

    1.1TB of HBM3e across 8 Hopper GPUs with 7.2TB/s of total aggregate bandwidth, nearly doubling the memory capacity and offering 1.4 times more memory bandwidth than HGX H100.

    Learn more
  • NVIDIA HGX H100

    Delivering exceptional performance, scalability, and security for every workload.

    Learn more