Together AI is a cloud platform for running, fine-tuning and scaling open-source AI models. Its pages describe inference, GPU clusters, storage, sandboxes and post-training under one roof.
The platform is built for developers and teams that want to use open models without managing their own infrastructure. Serverless Inference runs models on demand through APIs, with no infrastructure to set up and no long-term commitment. Batch Inference handles large asynchronous workloads. Provisioned Throughput offers committed capacity with a 99% uptime SLA, and Dedicated Model Inference runs models on dedicated hardware.
For teams that need more control, Accelerated Compute provides GPU clusters from instant self-serve setups up to thousands of GPUs. Managed Storage offers object storage and parallel filesystems with zero egress fees. Sandbox provides code environments for AI apps and agents.
For model work, Fine-Tuning launches managed supervised fine-tuning and DPO jobs on open-source models, and Post-Training lets teams define their own reinforcement learning and supervised fine-tuning loops. Evaluations measure model quality.
Pricing for serverless inference is per million tokens, with separate batch and cached rates listed for each model. Some models are listed as free. The pricing page says most teams start with serverless inference and move to dedicated endpoints as they grow, and it offers a contact form for help choosing a fit.
