Together AI

Together AI is a full-stack AI platform that offers inference, fine-tuning, and GPU cluster services for teams building with AI models. Its serverless inference lets users run open-source models on demand through an API, with no infrastructure to manage and no long-term commitments. For larger or steadier workloads, Together AI also offers batch inference, provisioned throughput with a stated 99% uptime SLA, and dedicated model and container inference on reserved hardware.

On the compute side, Together AI provides GPU clusters ranging from self-serve instant clusters to thousands of GPUs, along with managed storage and code sandboxes for building AI development environments. Its model shaping tools cover managed supervised fine-tuning and preference tuning jobs, as well as evaluations for measuring model quality. The company also publishes systems research, including work on kernels and inference, which it ties back to its production platform. The site lists a model library of open-source models across chat, image, code, vision, and video.